ccnet-devel Mailing List for CruiseControl.NET

+1 for Graham's idea on IntegrationStatus.UpdatingSource while calling
ISourceControl.GetSource

CM Synergy has the same issue here as ClearCase, as the queries that
populate the "view" can take awhile

That's a good point, I hadn't thought of that (we switched off labelling
because that takes another 5 minutes ;( )

 

I had second thoughts when I came to implement this change.  Adding lots
of new hard-coded values to the ProjectStatus enum seems to be travelling
in the opposite direction to the more generic approach to builders and
tasks that ccnet is now trying to take.

 

In fact, my real requirement is that the cctray icon goes yellow as soon
as possible once ccnet has decided to perform a build. So I raised and
implemented http://jira.public.thoughtworks.org/browse/CCNET-411 to move
the point at which the project status turns to building earlier: getting
the source and temporary labelling is included (just as it would have been
if using your own bootstrap nant script).

 

There's another "nice to have" feature here, which is that it'd be good to
have an idea of how far through a build ccnet is.  I haven't addressed
that.

 

g

 

 

  _____  

From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Bolles,
James E
Sent: 29 April 2005 2:34 pm
To: ccn...@li...
Subject: RE: [Ccnet-devel] Integration status during AutoGetSource

 

I like this idea and hope we can expand to add when items are being
Labeled as well.  Our Pvcs logic runs against source control servers in a
different state getting and labeling code does take a while.

 

Additional Status recommendations: Applying Label {0}, Applying Temporary
Label {0}

 

This should be an easy change for the IntegrationRunner.cs where this
status could be updated during the RunIntegration Method.

 

James

 

  _____  

From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Graham
Tackley
Sent: Friday, April 29, 2005 8:12 AM
To: ccn...@li...
Subject: [Ccnet-devel] Integration status during AutoGetSource

 

I've just committed an implementation of CCNET-409 - adding support for
"autoGetSource" to the ClearCase provider.

 

We've been testing it here over the last couple of weeks. The ClearCase
installation we are using is not the fastest of source control systems -
the update for one of our projects takes up to five minutes, depending on
server load.

 

While this is going on, the integration status is "CheckingModifications".
This is non-ideal as the cctray icon is still green and it's not obvious
that ccnet has decided that it's going to do a build.  For us, it stays
that way for five minutes, yuk.

 

I propose adding a new project activity status - UpdatingSource -- that
the integration goes into while calling ISourceControl.GetSource.  cctray
should go yellow during this state.

 

Comments/objections before I have a go at implementing this?

 

g

 

 

******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************

In the NAnt Documentation Example, it has buildTimeout element in the
full example but the configuration elements shows buildTimeoutSeconds
and uses that element.  Can we update this document on the website?

=20

Thanks, James

******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************

I like this idea and hope we can expand to add when items are being
Labeled as well.  Our Pvcs logic runs against source control servers in
a different state getting and labeling code does take a while.

=20

Additional Status recommendations: Applying Label {0}, Applying
Temporary Label {0}

=20

This should be an easy change for the IntegrationRunner.cs where this
status could be updated during the RunIntegration Method.

=20

James

=20

________________________________

From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Graham
Tackley
Sent: Friday, April 29, 2005 8:12 AM
To: ccn...@li...
Subject: [Ccnet-devel] Integration status during AutoGetSource

=20

I've just committed an implementation of CCNET-409 - adding support for
"autoGetSource" to the ClearCase provider.

=20

We've been testing it here over the last couple of weeks. The ClearCase
installation we are using is not the fastest of source control systems -
the update for one of our projects takes up to five minutes, depending
on server load.

=20

While this is going on, the integration status is
"CheckingModifications".  This is non-ideal as the cctray icon is still
green and it's not obvious that ccnet has decided that it's going to do
a build.  For us, it stays that way for five minutes, yuk.

=20

I propose adding a new project activity status - UpdatingSource -- that
the integration goes into while calling ISourceControl.GetSource.
cctray should go yellow during this state.

=20

Comments/objections before I have a go at implementing this?

=20

g

=20

=20

******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************

I've just committed an implementation of CCNET-409 - adding support for
"autoGetSource" to the ClearCase provider.

 

We've been testing it here over the last couple of weeks. The ClearCase
installation we are using is not the fastest of source control systems -
the update for one of our projects takes up to five minutes, depending on
server load.

 

While this is going on, the integration status is "CheckingModifications".
This is non-ideal as the cctray icon is still green and it's not obvious
that ccnet has decided that it's going to do a build.  For us, it stays
that way for five minutes, yuk.

 

I propose adding a new project activity status - UpdatingSource -- that
the integration goes into while calling ISourceControl.GetSource.  cctray
should go yellow during this state.

 

Comments/objections before I have a go at implementing this?

 

g

 

 

Here are the latest PVCS files to support labeling and gets.  I have
also improved the PvcsTest.cs and PvcsHistoryParser.cs as well.  Please
take a look and let me know if we need anything else. =20

I am unable to test logging into VM so I need someone to help test this.
Any takers?

James
=20
=20

-----Original Message-----
From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Bolles,
James E
Sent: Tuesday, April 26, 2005 5:34 PM
To: ccn...@li...
Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup work

I have just completed implementing the Labeling logic in Pvcs and I am
writing my test cases now.  If someone wants to help me test this logic
against a VM repository that requires login ids, it would help.  I was
hoping to implement the use of Promotion Groups in this cut but it needs
to hold of for more testing.  I should be done by Friday with updated
Test Cases and the final cut of the Pvcs and PvcsHistory changes.

About labeling in PVCS, I had to use Vcs -v to get the labels to be
capable of being applied in bulk.  If someone knows how to do this with
the pcli Label command with an input file (@filename), that would be
better!

Thanks,

James

=20
-----Original Message-----
From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Bolles,
James E
Sent: Monday, April 25, 2005 8:57 AM
To: ccn...@li...
Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup work

Thanks for the feedback, Owen.  I know you are busy so I will try to get
some of these items resolved ASAP while we have your attention :).  We
may have a work-around to not use the versioning at all going forward if
we use temporary labeling but I still feel that the Modification.cs
should support whatever versioning is needed by SourceControl providers.
Does it matter if it is a string or number?  We could just change it to
support strings and have only one format.

As for the PvcsHistoryTest, I do not see any such test in 0.9 under
core\sourcecontrol\test or core\test or UnitTest\core\SourceControl.  We
can add one and put together some different test at this level.  You had
said it is hanging indefinitely in an existing test.  Can you point me
in the correct direction to find this test and I am willing to fix it.

The PVCS test was updated to support the new inputs but, overall, I did
not add a lot of new tests.  I can do that as well. I will be looking at
this today and let you know why the test are hanging.  I have been
running this on my machine and have not experienced these results but
will try to replicate them here.

I am trying to get this all wrapped up because I would like to see this
get into a build on your side as soon as humanly possible. =20

Thanks,
James
=20
=20
-----Original Message-----
From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Owen
Rogers
Sent: Sunday, April 24, 2005 8:38 AM
To: ccn...@li...
Subject: Re: [Ccnet-devel] PVCS Files - Still needs some cleanup work

ok, i'm about a week late getting to this :-/
a few observations: i've done some simple code cleanup deleting unused
references and unused code.  i'm not overly happy about the addition
of the Version attribute to Modification, but i'll let it stand for
now.  we need to clean this up.  we might want to have separate
modification subclasses for different source controls.
aside from this, are you missing an updated version of
PvcsHistoryTest?  and are there some updated tests for the Pvcs class?
 it looks like one test in PvcsTest was deleted, but no tests were
added.
actually, now that i'm running the tests, the tests in PvcsHistoryTest
seem to hang indefinitely.  seems like:
_searchRegEx.Matches(ModificationFile) goes in but does not come back
out again.  i can't check this in until we've resolved this.  do you
know what might be the cause?
cheers,
owen.

On 4/16/05, Owen Rogers <exo...@gm...> wrote:
> jb and mike,
> i'll look into processing these patches this weekend.  i'm pretty much
> going to delegate to you two on design and testing as i currently
> don't have any way of testing this functionality.  so, don't expect
> much from me other than maybe some small code cleanup (if needed).
> btw, i'm enjoying the healthy debate around this on the mailing list.
> :)
> cheers,
> owen.
>=20
> On 4/15/05, Bolles, James E <Jam...@el...> wrote:
> > Here is the latest PVCS code with test cases.  Let me know what you
> > think.  The workspace is there but not in use yet.  The test cases
apply
> > to Vlog, Labeling, and the items that use to exist as well.  I have
> > tested this against a 7.5.1 VM instance and it works better than
before
> > with fewer instruction files as well.  Thanks, Mike, for the help.
> >
> > Owen or MR, what are the chances of getting someone on your side to
> > review this as well so we can get this into the build when we get
some
> > feedback?
> >
> > JB
> >
> > -----Original Message-----
> > From: ccn...@li...
> > [mailto:ccn...@li...] On Behalf Of
Michael
> > Luke
> > Sent: Thursday, April 14, 2005 11:00 AM
> > To: ccn...@li...
> > Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup
work
> >
> > Hi James.
> >
> > I can see what you've done now.  You're not relying upon the
labelling
> > to get the source.
> >
> > You are still only getting the files that have changed though, not
the
> > whole repository.  This could lead to problems with orphaned files
or
> > versioning issues.
> >
> > The way I work is to do the following:
> > - Nuke the working folder
> > - Label the latest version of all files in the projects
> > - Get all files with that label to the working folder
> > - Build
> > - Either delete the label (failed build) or rename the label to the
> > permanent one (successful build)
> >
> > I'm not sure I understand you when you say it might be a mistake to
get
> > the latest version.
> >
> > Mike
> >
> > Bolles, James E wrote:
> > > Mike, the reason we need version number is for getting the code
out of
> > > PVCS.  Without it, we would be pulling the latest code change and
that
> > > is not necessarily true.  Look at how I did the GetSource logic.
We
> > > need the version number or the gets will not work properly.  I
would
> > > like to expand PVCS to support using floating labels or promotion
> > > groups so we can label like VSS - maybe not this week, thought :)
> > >
> > > James E. Bolles
> > > El Paso Corporation
> > > Analyst - Information Technology - Sr
> > > Transportation & Gathering Business Systems - SNG
> > > 1900 5th Ave North, PO Box 2563
> > > Birmingham, AL  35203
> > > Work:  205.325.3502
> > > Fax:    205.326.2084
> > >
> > >
> > >
> > > -----Original Message-----
> > > From: ccn...@li...
> > > [mailto:ccn...@li...] On Behalf Of
Michael
> > > Luke
> > > Sent: Thursday, April 14, 2005 9:36 AM
> > > To: ccn...@li...
> > > Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup
work
> > >
> > > Inline:
> > >
> > > Bolles, James E wrote:
> > >> First of all, I don't take offense to coding comments.  My first
pass
> > >> is to get it to work, then get it to work well, followed by get
it to
> > >> look pretty :).  Here are my thoughts/responses:
> > >
> > > [ML] Thought that would be the case, but you can never be too
careful
> > > ;-)
> > >
> > >> Labeling:
> > >> - Labeling is of much use to us.  It allows us to determine what
> > >> files were successfully compiled with a build so we can
back-track
> > >> without the use of CC's xml files.  However, what I really want
is
> > >> to use a promotion group(PG) or base label so we can
promote/label
> > >> successfully built files to a particular PG/Label making it
easier
> > >> for the developer to get all source code for a product.  The
current
> > >> labeling method in place was done to mimic what VSS and MKS were
> > >> doing and I would like PVCS to function like the other
SourceControl
> > >> providers.
> > >
> > > [ML] Hmm.  Just had a look through the VSS code and it seems to
label
> > > ALL of the code, not just the files that have changed.  This is
the
> > > behaviour that makes most sense to me.
> > >
> > > I agree that being able to promote a successful build
automatically
> > > would be a good idea, but this is somewhat separate from
labelling,
> > > which is mainly used to ensure atomicity (label before, get label,
> > > build, rename label on completion).
> > >
> > >> VLog and ChangeNumber:
> > >> - Calling VLog directly does not fix the UNC problem.  Are PVCS
> > >> instances reside on servers so we have to get a list of the
versioned
> > >> files first, add the extra backslash, and then process the data.
I
> > >> will try to do the vlog as you described and I hope to have
results
> > >> by Friday.
> > >
> > > [ML] Yes.  If the earlier versions of PVCS support "pcli run vlog"
> > > then you will have no need to call listversionedfiles at all.
> > >
> > >> - As you know, PVCS doesn't really store version numbers as
numeric
> > >> but as strings.  If you do a branch, your version number will be
> > >> 1.x.y instead of just 1.x.  It may not be used but why is it in
> > >> Modification then?  Why not build the logic to support version
> > >> numbers from PVCS in the event it is added later?  I believe it
> > >> needs to be there.
> > >
> > > [ML] I've just tracked ChangeNumber through the source and it
appears
> > > to be being set to the greatest change number of the files in the
> > > modification list.  This is then written to the build log for
possible
> > > display by an XSLT.  As far as I can tell, ChangeNumber exists for
> > > those source control systems that uniquely identify checkins using
a
> > > numbering system.  PVCS, of course, doesn't do this.  The highest
> > > version number of a modified file is in no way analogous to the
> > > ChangeNumber of the last checkin in another source control system.
I
> > > don't think there is any benefit in knowing that one of the files
> > > just checked in was version
> > > 1.14 - do you?
> > >
> > >> Login:
> > >> - I have no idea how this works :(.  I do not have the ability to
> > >> test this functionality either because we are not allowed to use
> > >> anything but integrated authentication.
> > >> - It was a typo for me to have the "/" and it should have been a
":"
> > >> - Can you test this where you are?
> > >
> > > [ML] We do use this type of login, and it works happily in the
format
> > > -id"username":"password"
> > > I think this is the safest approach as it allows for spaces in
> > > username or password
> > >
> > >> UNC Support:
> > >> - I will add Martin's change in as well as look at his Test Case
as
> > >> well.
> > >
> > > [ML] Thanks.  Although you might not need if it we don't need to
call
> > > listversionedfiles
> > >
> > >> Code Review:
> > >> - I was not ready for code review as you have already guessed.  I
was
> > >> looking for feedback first and I appreciate what you have sent.
I
> > >> will clean up the code this week and resend it Friday.
> > >
> > > [ML] Great.  Sorry for being so blunt, but you can never be sure
just
> > > what the other person does and doesn't know.  I'll know for next
time!
> > >
> > >> - What would be the safest routine for cleaning up the files?  I
hate
> > >> using all of those temp files but I feel like we have to with the
way
> > >> PCLI works and I want to make sure they are all destroyed between
> > >> processes.  I will see what else we can do to make this safer.
> > >
> > > [ML] To be honest, I have a simple File.Delete in my code, without
any
> > > exception handling at all.  My reasoning being that it's better
for
> > > debugging when you can see what went wrong.  If you truly wanted
to
> > > remove the temporary files regardless, I'd put a try..catch in a
loop
> > > inside your finally that deleted each file one by one.  Just make
sure
> > > that you don't swallow any real exceptions, just ones caused by
the
> > > File.Delete failing.
> > >
> > > As for the temporary files - you really don't need that many.  All
of
> > > the Execute functions are set to use "pcli run -s" after you've
> > > generated a temporary file containing a single line PCLI script.
> > > Instead of creating this temporary file, just pass the single line
as
> > > the argument to the pcli.exe command.  See my example in the
> > > Merant::CreateTemporaryFile function.  You may need to refactor
the
> > > listversionedfiles command to make it a single line command rather
> > > than two lines.
> > >
> > >> JB
> > >>
> > >
> > > Mike
> >
> > -------------------------------------------------------
> > SF email is sponsored by - The IT Product Guide
> > Read honest & candid reviews on hundreds of IT Products from real
users.
> > Discover which products truly live up to the hype. Start reading
now.
> > http://ads.osdn.com/?ad_id=3D6595&alloc_id=3D14396&op=3Dclick
> > _______________________________________________
> > Ccnet-devel mailing list
> > Ccn...@li...
> > https://lists.sourceforge.net/lists/listinfo/ccnet-devel
> > ******************************************************************
> > This email and any files transmitted with it from the ElPaso
> > Corporation are confidential and intended solely for the
> > use of the individual or entity to whom they are addressed.
> > If you have received this email in error please notify the
> > sender.
> > ******************************************************************
> >
> >
> >
>=20
> --
> Owen Rogers | http://dotnetjunkies.com/weblog/exortech |
> CruiseControl.NET - http://ccnet.thoughtworks.com
>=20


--=20
Owen Rogers | http://dotnetjunkies.com/weblog/exortech |
CruiseControl.NET - http://ccnet.thoughtworks.com


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id396&op=3Dick
_______________________________________________
Ccnet-devel mailing list
Ccn...@li...
https://lists.sourceforge.net/lists/listinfo/ccnet-devel
******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id396&op=3Dick
_______________________________________________
Ccnet-devel mailing list
Ccn...@li...
https://lists.sourceforge.net/lists/listinfo/ccnet-devel
******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************


-------------------------------------------------------
SF.Net email is sponsored by: Tell us your software development plans!
Take this survey and enter to win a one-year sub to SourceForge.net
Plus IDC's 2005 look-ahead and a copy of this survey
Click here to start!  http://www.idcswdc.com/cgi-bin/survey?id5hix
_______________________________________________
Ccnet-devel mailing list
Ccn...@li...
https://lists.sourceforge.net/lists/listinfo/ccnet-devel
******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************

To add to you discussion - it doesnt matter whether you fill build and
revision numbers in assembly version with whatever number. It just make
assembly newer. Both numbers (ccnet build and svn revision) is =
increasing
and usually there is 1:1 corellation between them (except forced =
builds).

Only "problem" with such versioning is that framework distinguish two
assemblies with same major, same minor, and different build and/or =
revision.
That _is_ problem in some deployment scenarios (like GAC assembly =
patches).
AFAIK that should change in framework 2.0's fusion.


Other issue is, that currently its quite hard to label svn tree with =
those
labels. We should simplify this in future IMHO.

Regards,
Martin Aliger=20

> -----Original Message-----
> From: ccn...@li...=20
> [mailto:ccn...@li...] On Behalf Of=20
> Thomas Tomiczek
> Sent: Wednesday, April 27, 2005 10:57 AM
> To: ccn...@li...
> Subject: RE: [Ccnet-devel] Is build number available via=20
> IIntegrationResult?
>=20
> Actually, as a matter of fact, this is where the mixup happens.
>=20
> The "Build" number in the version is not what CC.NET=20
> understands as build number. Because within CC.NET a build=20
> can not have revisions.
>=20
> So, CC.NET - should increase the LAST number. The Revision.
>=20
> The Build is increased whenever you do a confirmed public=20
> bbuild - in our case this is daily, some proejcts weekly.=20
> These go up onto internal deployment directories.
>=20
> Thomas=20
>=20
> > -----Original Message-----
> > From: ccn...@li...
> > [mailto:ccn...@li...] On Behalf=20
> Of Marcin=20
> > Hoppe
> > Sent: Mittwoch, 27. April 2005 10:40
> > To: ccn...@li...
> > Subject: Re: [Ccnet-devel] Is build number available via=20
> > IIntegrationResult?
> >=20
> > On 4/27/05, Thomas Tomiczek <t.t...@th...> wrote:
> > > There is loss of functoinality, because the .NET loader behaves=20
> > > different depending on which order you put numbers in. This
> > is propery
> > > documented and gets important when multiple versions of=20
> an assembly=20
> > > are on the machine.
> >=20
> > I didn't have any problems with that (mainly due to lack of=20
> experience=20
> > in .NET deployment). However MSDN seems to prefer=20
> > major.minor[.build[.revision]] scheme (according to=20
> documentation of=20
> > System.Version class).
> >=20
> > According to my best knowledge and intuition: it is the build field=20
> > that should change with every CruiseControl.NET successful=20
> build. The=20
> > fact is, however, that this scheme is quite hard to achieve at the=20
> > moment. It requires a dirty trick in NAnt for example=20
> (adding manually=20
> > revision number).
> >=20
> > I could of course change sequence of build and revision=20
> fields -- and=20
> > that would simplify the building and labelling process. But=20
> what are=20
> > the side-effects regarding .NET Framework, I cannot tell,=20
> > unfortnately. That's why I use the "dirty" solution right now.
> >=20
> > What are your comments about this?
> >=20
> > Best regards!
> > Marcin
> >=20
> >=20
> > -------------------------------------------------------
> > SF.Net email is sponsored by: Tell us your software=20
> development plans!
> > Take this survey and enter to win a one-year sub to SourceForge.net=20
> > Plus IDC's 2005 look-ahead and a copy of this survey Click here to=20
> > start!
> > http://www.idcswdc.com/cgi-bin/survey?id=105hix
> > _______________________________________________
> > Ccnet-devel mailing list
> > Ccn...@li...
> > https://lists.sourceforge.net/lists/listinfo/ccnet-devel
> >=20
>=20
>=20
> -------------------------------------------------------
> SF.Net email is sponsored by: Tell us your software development plans!
> Take this survey and enter to win a one-year sub to=20
> SourceForge.net Plus IDC's 2005 look-ahead and a copy of this=20
> survey Click here to start! =20
> http://www.idcswdc.com/cgi-bin/survey?id=105hix
> _______________________________________________
> Ccnet-devel mailing list
> Ccn...@li...
> https://lists.sourceforge.net/lists/listinfo/ccnet-devel
>=20

On 4/27/05, Thomas Tomiczek <t.t...@th...> wrote:
> Actually, as a matter of fact, this is where the mixup happens.
>=20
> The "Build" number in the version is not what CC.NET understands as
> build number. Because within CC.NET a build can not have revisions.
>=20
> So, CC.NET - should increase the LAST number. The Revision.
>=20
> The Build is increased whenever you do a confirmed public bbuild - in
> our case this is daily, some proejcts weekly. These go up onto internal
> deployment directories.

Right you are, Thomas.

Therefore it is desirable that CruiseControl.NET give developers means
to somehow increase both numbers in isolation (to some extent, of
course).

What comes to my mind right now is:
 * automatic integration builds should increase last (revision) field
-- that should be one trigger,
 * daily/weekly/interal deployment build should increase last but one
(build) field and reset the revision fields -- that is second trigger.

That yields the need for:
 * naming triggers (is it under development or under discussion right now?)=
,
 * far more flexible labellers (the above question remains).

Best regards!
Marcin

Actually, as a matter of fact, this is where the mixup happens.

The "Build" number in the version is not what CC.NET understands as
build number. Because within CC.NET a build can not have revisions.

So, CC.NET - should increase the LAST number. The Revision.

The Build is increased whenever you do a confirmed public bbuild - in
our case this is daily, some proejcts weekly. These go up onto internal
deployment directories.

Thomas=20

> -----Original Message-----
> From: ccn...@li...=20
> [mailto:ccn...@li...] On Behalf Of=20
> Marcin Hoppe
> Sent: Mittwoch, 27. April 2005 10:40
> To: ccn...@li...
> Subject: Re: [Ccnet-devel] Is build number available via=20
> IIntegrationResult?
>=20
> On 4/27/05, Thomas Tomiczek <t.t...@th...> wrote:
> > There is loss of functoinality, because the .NET loader behaves=20
> > different depending on which order you put numbers in. This=20
> is propery=20
> > documented and gets important when multiple versions of an assembly=20
> > are on the machine.
>=20
> I didn't have any problems with that (mainly due to lack of=20
> experience in .NET deployment). However MSDN seems to prefer=20
> major.minor[.build[.revision]] scheme (according to=20
> documentation of System.Version class).
>=20
> According to my best knowledge and intuition: it is the build=20
> field that should change with every CruiseControl.NET=20
> successful build. The fact is, however, that this scheme is=20
> quite hard to achieve at the moment. It requires a dirty=20
> trick in NAnt for example (adding manually revision number).
>=20
> I could of course change sequence of build and revision=20
> fields -- and that would simplify the building and labelling=20
> process. But what are the side-effects regarding .NET=20
> Framework, I cannot tell, unfortnately. That's why I use the=20
> "dirty" solution right now.
>=20
> What are your comments about this?
>=20
> Best regards!
> Marcin
>=20
>=20
> -------------------------------------------------------
> SF.Net email is sponsored by: Tell us your software development plans!
> Take this survey and enter to win a one-year sub to=20
> SourceForge.net Plus IDC's 2005 look-ahead and a copy of this=20
> survey Click here to start! =20
> http://www.idcswdc.com/cgi-bin/survey?id=105hix
> _______________________________________________
> Ccnet-devel mailing list
> Ccn...@li...
> https://lists.sourceforge.net/lists/listinfo/ccnet-devel
>=20

On 4/27/05, Thomas Tomiczek <t.t...@th...> wrote:
> There is loss of functoinality, because the .NET loader behaves
> different depending on which order you put numbers in. This is propery
> documented and gets important when multiple versions of an assembly are
> on the machine.

I didn't have any problems with that (mainly due to lack of experience
in .NET deployment). However MSDN seems to prefer
major.minor[.build[.revision]] scheme (according to documentation of
System.Version class).

According to my best knowledge and intuition: it is the build field
that should change with every CruiseControl.NET successful build. The
fact is, however, that this scheme is quite hard to achieve at the
moment. It requires a dirty trick in NAnt for example (adding manually
revision number).

I could of course change sequence of build and revision fields -- and
that would simplify the building and labelling process. But what are
the side-effects regarding .NET Framework, I cannot tell,
unfortnately. That's why I use the "dirty" solution right now.

What are your comments about this?

Best regards!
Marcin

There is loss of functoinality, because the .NET loader behaves
different depending on which order you put numbers in. This is propery
documented and gets important when multiple versions of an assembly are
on the machine.

Thomas=20

> -----Original Message-----
> From: ccn...@li...=20
> [mailto:ccn...@li...] On Behalf Of=20
> Marcin Hoppe
> Sent: Mittwoch, 27. April 2005 10:21
> To: ccn...@li...
> Subject: Re: [Ccnet-devel] Is build number available via=20
> IIntegrationResult?
>=20
> On 4/27/05, Thomas Tomiczek <t.t...@th...> wrote:
> > Marcin, excuse me to jump in . This is not the first time I=20
> see your=20
> > numbering scheme.
>=20
> Yes. It's fairly standard, I think.
>=20
> > Just, unless you redefine Revision and build - how do you=20
> justify the=20
> > loss of functionality of schanging numbers? It is not=20
> exactly like the=20
> > .NET loader is not changing it's behavior just because you redefine=20
> > order of the numbers.
>=20
> I must admit that I don't understand your question -- what do=20
> you mean by "loss of functionality"?
>=20
> There's no loss of functionality in my opinion. I've been=20
> just wondering if I can accustom CruiseControl.NET somehow to=20
> follow scheme that I'm using.
>=20
> Best regards!
> Marcin
>=20
>=20
> -------------------------------------------------------
> SF.Net email is sponsored by: Tell us your software development plans!
> Take this survey and enter to win a one-year sub to=20
> SourceForge.net Plus IDC's 2005 look-ahead and a copy of this=20
> survey Click here to start! =20
> http://www.idcswdc.com/cgi-bin/survey?id=105hix
> _______________________________________________
> Ccnet-devel mailing list
> Ccn...@li...
> https://lists.sourceforge.net/lists/listinfo/ccnet-devel
>=20

On 4/27/05, Thomas Tomiczek <t.t...@th...> wrote:
> Marcin, excuse me to jump in . This is not the first time I see your
> numbering scheme.

Yes. It's fairly standard, I think.

> Just, unless you redefine Revision and build - how do you justify the
> loss of functionality of schanging numbers? It is not exactly like the
> .NET loader is not changing it's behavior just because you redefine
> order of the numbers.

I must admit that I don't understand your question -- what do you mean
by "loss of functionality"?

There's no loss of functionality in my opinion. I've been just
wondering if I can accustom CruiseControl.NET somehow to follow scheme
that I'm using.

Best regards!
Marcin

Marcin, excuse me to jump in . This is not the first time I see your
numbering scheme.

Just, unless you redefine Revision and build - how do you justify the
loss of functionality of schanging numbers? It is not exactly like the
.NET loader is not changing it's behavior just because you redefine
order of the numbers.

Thomas=20

> -----Original Message-----
> From: ccn...@li...=20
> [mailto:ccn...@li...] On Behalf Of=20
> Marcin Hoppe
> Sent: Mittwoch, 27. April 2005 09:51
> To: ccn...@li...
> Subject: Re: [Ccnet-devel] Is build number available via=20
> IIntegrationResult?
>=20
> On 4/25/05, Jeff Brown <Je...@in...> wrote:
> > I'll second that.
> > We would like to label builds using the SVN revision number=20
> as one of=20
> > the components.
> > One scheme we're considering: major.minor.revision.build#.
>=20
> Yup. Unfortunately it's different from the scheme I'm using:
> major.minor.build.revision (which is more or less a standard,=20
> I guess).
>=20
> > It would be nifty to have a labeller implementation that allowed=20
> > substitution strings.  We could expose a number of "interesting"
> > properties of a build that can be combined to form the label.  This=20
> > subsumes the notion of prefix and suffix.
>=20
> Sure, it would be nice. Is there any progress on this?
>=20
> Best regards!
> Marcin
>=20
>=20
> -------------------------------------------------------
> SF.Net email is sponsored by: Tell us your software development plans!
> Take this survey and enter to win a one-year sub to=20
> SourceForge.net Plus IDC's 2005 look-ahead and a copy of this=20
> survey Click here to start! =20
> http://www.idcswdc.com/cgi-bin/survey?id=105hix
> _______________________________________________
> Ccnet-devel mailing list
> Ccn...@li...
> https://lists.sourceforge.net/lists/listinfo/ccnet-devel
>=20

On 4/25/05, Jeff Brown <Je...@in...> wrote:
> I'll second that.
> We would like to label builds using the SVN revision number as one of
> the components.
> One scheme we're considering: major.minor.revision.build#.

Yup. Unfortunately it's different from the scheme I'm using:
major.minor.build.revision (which is more or less a standard, I
guess).

> It would be nifty to have a labeller implementation that allowed
> substitution strings.  We could expose a number of "interesting"
> properties of a build that can be combined to form the label.  This
> subsumes the notion of prefix and suffix.

Sure, it would be nice. Is there any progress on this?

Best regards!
Marcin

Hi,

Also check out the related jira issue 399.


Regards,

Freek.

I have just completed implementing the Labeling logic in Pvcs and I am
writing my test cases now.  If someone wants to help me test this logic
against a VM repository that requires login ids, it would help.  I was
hoping to implement the use of Promotion Groups in this cut but it needs
to hold of for more testing.  I should be done by Friday with updated
Test Cases and the final cut of the Pvcs and PvcsHistory changes.

About labeling in PVCS, I had to use Vcs -v to get the labels to be
capable of being applied in bulk.  If someone knows how to do this with
the pcli Label command with an input file (@filename), that would be
better!

Thanks,

James

=20
-----Original Message-----
From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Bolles,
James E
Sent: Monday, April 25, 2005 8:57 AM
To: ccn...@li...
Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup work

Thanks for the feedback, Owen.  I know you are busy so I will try to get
some of these items resolved ASAP while we have your attention :).  We
may have a work-around to not use the versioning at all going forward if
we use temporary labeling but I still feel that the Modification.cs
should support whatever versioning is needed by SourceControl providers.
Does it matter if it is a string or number?  We could just change it to
support strings and have only one format.

As for the PvcsHistoryTest, I do not see any such test in 0.9 under
core\sourcecontrol\test or core\test or UnitTest\core\SourceControl.  We
can add one and put together some different test at this level.  You had
said it is hanging indefinitely in an existing test.  Can you point me
in the correct direction to find this test and I am willing to fix it.

The PVCS test was updated to support the new inputs but, overall, I did
not add a lot of new tests.  I can do that as well. I will be looking at
this today and let you know why the test are hanging.  I have been
running this on my machine and have not experienced these results but
will try to replicate them here.

I am trying to get this all wrapped up because I would like to see this
get into a build on your side as soon as humanly possible. =20

Thanks,
James
=20
=20
-----Original Message-----
From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Owen
Rogers
Sent: Sunday, April 24, 2005 8:38 AM
To: ccn...@li...
Subject: Re: [Ccnet-devel] PVCS Files - Still needs some cleanup work

ok, i'm about a week late getting to this :-/
a few observations: i've done some simple code cleanup deleting unused
references and unused code.  i'm not overly happy about the addition
of the Version attribute to Modification, but i'll let it stand for
now.  we need to clean this up.  we might want to have separate
modification subclasses for different source controls.
aside from this, are you missing an updated version of
PvcsHistoryTest?  and are there some updated tests for the Pvcs class?
 it looks like one test in PvcsTest was deleted, but no tests were
added.
actually, now that i'm running the tests, the tests in PvcsHistoryTest
seem to hang indefinitely.  seems like:
_searchRegEx.Matches(ModificationFile) goes in but does not come back
out again.  i can't check this in until we've resolved this.  do you
know what might be the cause?
cheers,
owen.

On 4/16/05, Owen Rogers <exo...@gm...> wrote:
> jb and mike,
> i'll look into processing these patches this weekend.  i'm pretty much
> going to delegate to you two on design and testing as i currently
> don't have any way of testing this functionality.  so, don't expect
> much from me other than maybe some small code cleanup (if needed).
> btw, i'm enjoying the healthy debate around this on the mailing list.
> :)
> cheers,
> owen.
>=20
> On 4/15/05, Bolles, James E <Jam...@el...> wrote:
> > Here is the latest PVCS code with test cases.  Let me know what you
> > think.  The workspace is there but not in use yet.  The test cases
apply
> > to Vlog, Labeling, and the items that use to exist as well.  I have
> > tested this against a 7.5.1 VM instance and it works better than
before
> > with fewer instruction files as well.  Thanks, Mike, for the help.
> >
> > Owen or MR, what are the chances of getting someone on your side to
> > review this as well so we can get this into the build when we get
some
> > feedback?
> >
> > JB
> >
> > -----Original Message-----
> > From: ccn...@li...
> > [mailto:ccn...@li...] On Behalf Of
Michael
> > Luke
> > Sent: Thursday, April 14, 2005 11:00 AM
> > To: ccn...@li...
> > Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup
work
> >
> > Hi James.
> >
> > I can see what you've done now.  You're not relying upon the
labelling
> > to get the source.
> >
> > You are still only getting the files that have changed though, not
the
> > whole repository.  This could lead to problems with orphaned files
or
> > versioning issues.
> >
> > The way I work is to do the following:
> > - Nuke the working folder
> > - Label the latest version of all files in the projects
> > - Get all files with that label to the working folder
> > - Build
> > - Either delete the label (failed build) or rename the label to the
> > permanent one (successful build)
> >
> > I'm not sure I understand you when you say it might be a mistake to
get
> > the latest version.
> >
> > Mike
> >
> > Bolles, James E wrote:
> > > Mike, the reason we need version number is for getting the code
out of
> > > PVCS.  Without it, we would be pulling the latest code change and
that
> > > is not necessarily true.  Look at how I did the GetSource logic.
We
> > > need the version number or the gets will not work properly.  I
would
> > > like to expand PVCS to support using floating labels or promotion
> > > groups so we can label like VSS - maybe not this week, thought :)
> > >
> > > James E. Bolles
> > > El Paso Corporation
> > > Analyst - Information Technology - Sr
> > > Transportation & Gathering Business Systems - SNG
> > > 1900 5th Ave North, PO Box 2563
> > > Birmingham, AL  35203
> > > Work:  205.325.3502
> > > Fax:    205.326.2084
> > >
> > >
> > >
> > > -----Original Message-----
> > > From: ccn...@li...
> > > [mailto:ccn...@li...] On Behalf Of
Michael
> > > Luke
> > > Sent: Thursday, April 14, 2005 9:36 AM
> > > To: ccn...@li...
> > > Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup
work
> > >
> > > Inline:
> > >
> > > Bolles, James E wrote:
> > >> First of all, I don't take offense to coding comments.  My first
pass
> > >> is to get it to work, then get it to work well, followed by get
it to
> > >> look pretty :).  Here are my thoughts/responses:
> > >
> > > [ML] Thought that would be the case, but you can never be too
careful
> > > ;-)
> > >
> > >> Labeling:
> > >> - Labeling is of much use to us.  It allows us to determine what
> > >> files were successfully compiled with a build so we can
back-track
> > >> without the use of CC's xml files.  However, what I really want
is
> > >> to use a promotion group(PG) or base label so we can
promote/label
> > >> successfully built files to a particular PG/Label making it
easier
> > >> for the developer to get all source code for a product.  The
current
> > >> labeling method in place was done to mimic what VSS and MKS were
> > >> doing and I would like PVCS to function like the other
SourceControl
> > >> providers.
> > >
> > > [ML] Hmm.  Just had a look through the VSS code and it seems to
label
> > > ALL of the code, not just the files that have changed.  This is
the
> > > behaviour that makes most sense to me.
> > >
> > > I agree that being able to promote a successful build
automatically
> > > would be a good idea, but this is somewhat separate from
labelling,
> > > which is mainly used to ensure atomicity (label before, get label,
> > > build, rename label on completion).
> > >
> > >> VLog and ChangeNumber:
> > >> - Calling VLog directly does not fix the UNC problem.  Are PVCS
> > >> instances reside on servers so we have to get a list of the
versioned
> > >> files first, add the extra backslash, and then process the data.
I
> > >> will try to do the vlog as you described and I hope to have
results
> > >> by Friday.
> > >
> > > [ML] Yes.  If the earlier versions of PVCS support "pcli run vlog"
> > > then you will have no need to call listversionedfiles at all.
> > >
> > >> - As you know, PVCS doesn't really store version numbers as
numeric
> > >> but as strings.  If you do a branch, your version number will be
> > >> 1.x.y instead of just 1.x.  It may not be used but why is it in
> > >> Modification then?  Why not build the logic to support version
> > >> numbers from PVCS in the event it is added later?  I believe it
> > >> needs to be there.
> > >
> > > [ML] I've just tracked ChangeNumber through the source and it
appears
> > > to be being set to the greatest change number of the files in the
> > > modification list.  This is then written to the build log for
possible
> > > display by an XSLT.  As far as I can tell, ChangeNumber exists for
> > > those source control systems that uniquely identify checkins using
a
> > > numbering system.  PVCS, of course, doesn't do this.  The highest
> > > version number of a modified file is in no way analogous to the
> > > ChangeNumber of the last checkin in another source control system.
I
> > > don't think there is any benefit in knowing that one of the files
> > > just checked in was version
> > > 1.14 - do you?
> > >
> > >> Login:
> > >> - I have no idea how this works :(.  I do not have the ability to
> > >> test this functionality either because we are not allowed to use
> > >> anything but integrated authentication.
> > >> - It was a typo for me to have the "/" and it should have been a
":"
> > >> - Can you test this where you are?
> > >
> > > [ML] We do use this type of login, and it works happily in the
format
> > > -id"username":"password"
> > > I think this is the safest approach as it allows for spaces in
> > > username or password
> > >
> > >> UNC Support:
> > >> - I will add Martin's change in as well as look at his Test Case
as
> > >> well.
> > >
> > > [ML] Thanks.  Although you might not need if it we don't need to
call
> > > listversionedfiles
> > >
> > >> Code Review:
> > >> - I was not ready for code review as you have already guessed.  I
was
> > >> looking for feedback first and I appreciate what you have sent.
I
> > >> will clean up the code this week and resend it Friday.
> > >
> > > [ML] Great.  Sorry for being so blunt, but you can never be sure
just
> > > what the other person does and doesn't know.  I'll know for next
time!
> > >
> > >> - What would be the safest routine for cleaning up the files?  I
hate
> > >> using all of those temp files but I feel like we have to with the
way
> > >> PCLI works and I want to make sure they are all destroyed between
> > >> processes.  I will see what else we can do to make this safer.
> > >
> > > [ML] To be honest, I have a simple File.Delete in my code, without
any
> > > exception handling at all.  My reasoning being that it's better
for
> > > debugging when you can see what went wrong.  If you truly wanted
to
> > > remove the temporary files regardless, I'd put a try..catch in a
loop
> > > inside your finally that deleted each file one by one.  Just make
sure
> > > that you don't swallow any real exceptions, just ones caused by
the
> > > File.Delete failing.
> > >
> > > As for the temporary files - you really don't need that many.  All
of
> > > the Execute functions are set to use "pcli run -s" after you've
> > > generated a temporary file containing a single line PCLI script.
> > > Instead of creating this temporary file, just pass the single line
as
> > > the argument to the pcli.exe command.  See my example in the
> > > Merant::CreateTemporaryFile function.  You may need to refactor
the
> > > listversionedfiles command to make it a single line command rather
> > > than two lines.
> > >
> > >> JB
> > >>
> > >
> > > Mike
> >
> > -------------------------------------------------------
> > SF email is sponsored by - The IT Product Guide
> > Read honest & candid reviews on hundreds of IT Products from real
users.
> > Discover which products truly live up to the hype. Start reading
now.
> > http://ads.osdn.com/?ad_id=3D6595&alloc_id=3D14396&op=3Dclick
> > _______________________________________________
> > Ccnet-devel mailing list
> > Ccn...@li...
> > https://lists.sourceforge.net/lists/listinfo/ccnet-devel
> > ******************************************************************
> > This email and any files transmitted with it from the ElPaso
> > Corporation are confidential and intended solely for the
> > use of the individual or entity to whom they are addressed.
> > If you have received this email in error please notify the
> > sender.
> > ******************************************************************
> >
> >
> >
>=20
> --
> Owen Rogers | http://dotnetjunkies.com/weblog/exortech |
> CruiseControl.NET - http://ccnet.thoughtworks.com
>=20


--=20
Owen Rogers | http://dotnetjunkies.com/weblog/exortech |
CruiseControl.NET - http://ccnet.thoughtworks.com


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id396&op=3Dick
_______________________________________________
Ccnet-devel mailing list
Ccn...@li...
https://lists.sourceforge.net/lists/listinfo/ccnet-devel
******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id396&op=3Dick
_______________________________________________
Ccnet-devel mailing list
Ccn...@li...
https://lists.sourceforge.net/lists/listinfo/ccnet-devel
******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************

On 4/25/05, Foster, Richard - PAL <RF...@qu...> wrote:

> P.P.S. I've been meaning to mention this for some time. I think there's
> a typo on
> http://confluence.public.thoughtworks.org/display/CCNET/Project+Configur
> ation+Block - Surely it should be "artifactDirectory" not "
> artifcactDirectory"?

Fixed - thanks.

Mike

I'll second that.
We would like to label builds using the SVN revision number as one of
the components.
One scheme we're considering: major.minor.revision.build#.

It would be nifty to have a labeller implementation that allowed
substitution strings.  We could expose a number of "interesting"
properties of a build that can be combined to form the label.  This
subsumes the notion of prefix and suffix.

Jeff.

-----Original Message-----
From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Marcin
Hoppe
Sent: Monday, April 25, 2005 2:30 AM
To: ccn...@li...
Subject: Re: [Ccnet-devel] Is build number available via
IIntegrationResult?

On 4/24/05, Owen Rogers <exo...@gm...> wrote:
> basically what this says is that it will match and replace any numbers
> (\d+) appearring at the end of the string ($) coming after an optional

> set of characters (the prefix: .*?)

Ouch! It's not feasible to use this with any suffix feature, I guess.

> not exactly.  we currently only pass around the label which contains=20
> the build number -- though it has to be parsed out.  what i would=20
> prefer to do (which is what i was going to do in the original=20
> implementation until YAGNI kicked in) is to introduce an ILabel=20
> interface that would enable you to retrieve this kind of information
> -- without having to rely on parsing.  Labellers basically would serve

> as a factory for a specific type of ILabel.

Do you plan to do it or is the idea dead? I like this idea and I'll try
to dig a little bit deeper in the codebase and take a look how
complicated this might be (I don't have the faintest idea at the
moment).

> this is used by the Subversion history parser.  there has been some=20
> discussion on this list to use the change set number as the build=20
> number.  this would keep ccnet and svn in sync.  we really should have

> a ModificationSet object that could encapsulate this.

Which is not always good, IMHO. This idea won't work properly when there
are several projects in the same repository (misterious gaps will
appear). However, such a feature may be desirable for some developers...
It is an ideal candidate for a new labeller for me. I have probably
missed sime discussions about it :).

> > Is it difficult
> > to modify the system to store build number in IIntegrationResult=20
> > explicitly?
>=20
> nope.  but i would prefer to encapsulate this in an object rather than

> continuing to expand the IntegrationResult interface.

Sure :). I'll try to investigate the topic while you're in the Himalayas
:).

Best regards!
Marcin


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide Read honest & candid
reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id=14396&op=3Dick
_______________________________________________
Ccnet-devel mailing list
Ccn...@li...
https://lists.sourceforge.net/lists/listinfo/ccnet-devel

Owen asked:
out of curiousity, when you do a "get latest"/update from coherence,
does it delete files from your local workspace if those files have been
deleted from the repository?  vss doesn't do this, which is why they do
some clever stuff in the nant <vss> task to clean up these files.  no
big deal, but i was wondering if you would have to do something similar
to get this to work.

I tested this with the GUI interface to Team Coherence and the "deleted"
object does not get erased.  I've avoided this in my manual nant builds
by starting the build with an empty source folder.


Stephen Lewis
Envisioneering LLC - Medical Products
St. Louis, MO 63114
(314) 429-7367 x112
=20

Thanks for the feedback, Owen.  I know you are busy so I will try to get
some of these items resolved ASAP while we have your attention :).  We
may have a work-around to not use the versioning at all going forward if
we use temporary labeling but I still feel that the Modification.cs
should support whatever versioning is needed by SourceControl providers.
Does it matter if it is a string or number?  We could just change it to
support strings and have only one format.

As for the PvcsHistoryTest, I do not see any such test in 0.9 under
core\sourcecontrol\test or core\test or UnitTest\core\SourceControl.  We
can add one and put together some different test at this level.  You had
said it is hanging indefinitely in an existing test.  Can you point me
in the correct direction to find this test and I am willing to fix it.

The PVCS test was updated to support the new inputs but, overall, I did
not add a lot of new tests.  I can do that as well. I will be looking at
this today and let you know why the test are hanging.  I have been
running this on my machine and have not experienced these results but
will try to replicate them here.

I am trying to get this all wrapped up because I would like to see this
get into a build on your side as soon as humanly possible. =20

Thanks,
James
=20
=20
-----Original Message-----
From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Owen
Rogers
Sent: Sunday, April 24, 2005 8:38 AM
To: ccn...@li...
Subject: Re: [Ccnet-devel] PVCS Files - Still needs some cleanup work

ok, i'm about a week late getting to this :-/
a few observations: i've done some simple code cleanup deleting unused
references and unused code.  i'm not overly happy about the addition
of the Version attribute to Modification, but i'll let it stand for
now.  we need to clean this up.  we might want to have separate
modification subclasses for different source controls.
aside from this, are you missing an updated version of
PvcsHistoryTest?  and are there some updated tests for the Pvcs class?
 it looks like one test in PvcsTest was deleted, but no tests were
added.
actually, now that i'm running the tests, the tests in PvcsHistoryTest
seem to hang indefinitely.  seems like:
_searchRegEx.Matches(ModificationFile) goes in but does not come back
out again.  i can't check this in until we've resolved this.  do you
know what might be the cause?
cheers,
owen.

On 4/16/05, Owen Rogers <exo...@gm...> wrote:
> jb and mike,
> i'll look into processing these patches this weekend.  i'm pretty much
> going to delegate to you two on design and testing as i currently
> don't have any way of testing this functionality.  so, don't expect
> much from me other than maybe some small code cleanup (if needed).
> btw, i'm enjoying the healthy debate around this on the mailing list.
> :)
> cheers,
> owen.
>=20
> On 4/15/05, Bolles, James E <Jam...@el...> wrote:
> > Here is the latest PVCS code with test cases.  Let me know what you
> > think.  The workspace is there but not in use yet.  The test cases
apply
> > to Vlog, Labeling, and the items that use to exist as well.  I have
> > tested this against a 7.5.1 VM instance and it works better than
before
> > with fewer instruction files as well.  Thanks, Mike, for the help.
> >
> > Owen or MR, what are the chances of getting someone on your side to
> > review this as well so we can get this into the build when we get
some
> > feedback?
> >
> > JB
> >
> > -----Original Message-----
> > From: ccn...@li...
> > [mailto:ccn...@li...] On Behalf Of
Michael
> > Luke
> > Sent: Thursday, April 14, 2005 11:00 AM
> > To: ccn...@li...
> > Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup
work
> >
> > Hi James.
> >
> > I can see what you've done now.  You're not relying upon the
labelling
> > to get the source.
> >
> > You are still only getting the files that have changed though, not
the
> > whole repository.  This could lead to problems with orphaned files
or
> > versioning issues.
> >
> > The way I work is to do the following:
> > - Nuke the working folder
> > - Label the latest version of all files in the projects
> > - Get all files with that label to the working folder
> > - Build
> > - Either delete the label (failed build) or rename the label to the
> > permanent one (successful build)
> >
> > I'm not sure I understand you when you say it might be a mistake to
get
> > the latest version.
> >
> > Mike
> >
> > Bolles, James E wrote:
> > > Mike, the reason we need version number is for getting the code
out of
> > > PVCS.  Without it, we would be pulling the latest code change and
that
> > > is not necessarily true.  Look at how I did the GetSource logic.
We
> > > need the version number or the gets will not work properly.  I
would
> > > like to expand PVCS to support using floating labels or promotion
> > > groups so we can label like VSS - maybe not this week, thought :)
> > >
> > > James E. Bolles
> > > El Paso Corporation
> > > Analyst - Information Technology - Sr
> > > Transportation & Gathering Business Systems - SNG
> > > 1900 5th Ave North, PO Box 2563
> > > Birmingham, AL  35203
> > > Work:  205.325.3502
> > > Fax:    205.326.2084
> > >
> > >
> > >
> > > -----Original Message-----
> > > From: ccn...@li...
> > > [mailto:ccn...@li...] On Behalf Of
Michael
> > > Luke
> > > Sent: Thursday, April 14, 2005 9:36 AM
> > > To: ccn...@li...
> > > Subject: RE: [Ccnet-devel] PVCS Files - Still needs some cleanup
work
> > >
> > > Inline:
> > >
> > > Bolles, James E wrote:
> > >> First of all, I don't take offense to coding comments.  My first
pass
> > >> is to get it to work, then get it to work well, followed by get
it to
> > >> look pretty :).  Here are my thoughts/responses:
> > >
> > > [ML] Thought that would be the case, but you can never be too
careful
> > > ;-)
> > >
> > >> Labeling:
> > >> - Labeling is of much use to us.  It allows us to determine what
> > >> files were successfully compiled with a build so we can
back-track
> > >> without the use of CC's xml files.  However, what I really want
is
> > >> to use a promotion group(PG) or base label so we can
promote/label
> > >> successfully built files to a particular PG/Label making it
easier
> > >> for the developer to get all source code for a product.  The
current
> > >> labeling method in place was done to mimic what VSS and MKS were
> > >> doing and I would like PVCS to function like the other
SourceControl
> > >> providers.
> > >
> > > [ML] Hmm.  Just had a look through the VSS code and it seems to
label
> > > ALL of the code, not just the files that have changed.  This is
the
> > > behaviour that makes most sense to me.
> > >
> > > I agree that being able to promote a successful build
automatically
> > > would be a good idea, but this is somewhat separate from
labelling,
> > > which is mainly used to ensure atomicity (label before, get label,
> > > build, rename label on completion).
> > >
> > >> VLog and ChangeNumber:
> > >> - Calling VLog directly does not fix the UNC problem.  Are PVCS
> > >> instances reside on servers so we have to get a list of the
versioned
> > >> files first, add the extra backslash, and then process the data.
I
> > >> will try to do the vlog as you described and I hope to have
results
> > >> by Friday.
> > >
> > > [ML] Yes.  If the earlier versions of PVCS support "pcli run vlog"
> > > then you will have no need to call listversionedfiles at all.
> > >
> > >> - As you know, PVCS doesn't really store version numbers as
numeric
> > >> but as strings.  If you do a branch, your version number will be
> > >> 1.x.y instead of just 1.x.  It may not be used but why is it in
> > >> Modification then?  Why not build the logic to support version
> > >> numbers from PVCS in the event it is added later?  I believe it
> > >> needs to be there.
> > >
> > > [ML] I've just tracked ChangeNumber through the source and it
appears
> > > to be being set to the greatest change number of the files in the
> > > modification list.  This is then written to the build log for
possible
> > > display by an XSLT.  As far as I can tell, ChangeNumber exists for
> > > those source control systems that uniquely identify checkins using
a
> > > numbering system.  PVCS, of course, doesn't do this.  The highest
> > > version number of a modified file is in no way analogous to the
> > > ChangeNumber of the last checkin in another source control system.
I
> > > don't think there is any benefit in knowing that one of the files
> > > just checked in was version
> > > 1.14 - do you?
> > >
> > >> Login:
> > >> - I have no idea how this works :(.  I do not have the ability to
> > >> test this functionality either because we are not allowed to use
> > >> anything but integrated authentication.
> > >> - It was a typo for me to have the "/" and it should have been a
":"
> > >> - Can you test this where you are?
> > >
> > > [ML] We do use this type of login, and it works happily in the
format
> > > -id"username":"password"
> > > I think this is the safest approach as it allows for spaces in
> > > username or password
> > >
> > >> UNC Support:
> > >> - I will add Martin's change in as well as look at his Test Case
as
> > >> well.
> > >
> > > [ML] Thanks.  Although you might not need if it we don't need to
call
> > > listversionedfiles
> > >
> > >> Code Review:
> > >> - I was not ready for code review as you have already guessed.  I
was
> > >> looking for feedback first and I appreciate what you have sent.
I
> > >> will clean up the code this week and resend it Friday.
> > >
> > > [ML] Great.  Sorry for being so blunt, but you can never be sure
just
> > > what the other person does and doesn't know.  I'll know for next
time!
> > >
> > >> - What would be the safest routine for cleaning up the files?  I
hate
> > >> using all of those temp files but I feel like we have to with the
way
> > >> PCLI works and I want to make sure they are all destroyed between
> > >> processes.  I will see what else we can do to make this safer.
> > >
> > > [ML] To be honest, I have a simple File.Delete in my code, without
any
> > > exception handling at all.  My reasoning being that it's better
for
> > > debugging when you can see what went wrong.  If you truly wanted
to
> > > remove the temporary files regardless, I'd put a try..catch in a
loop
> > > inside your finally that deleted each file one by one.  Just make
sure
> > > that you don't swallow any real exceptions, just ones caused by
the
> > > File.Delete failing.
> > >
> > > As for the temporary files - you really don't need that many.  All
of
> > > the Execute functions are set to use "pcli run -s" after you've
> > > generated a temporary file containing a single line PCLI script.
> > > Instead of creating this temporary file, just pass the single line
as
> > > the argument to the pcli.exe command.  See my example in the
> > > Merant::CreateTemporaryFile function.  You may need to refactor
the
> > > listversionedfiles command to make it a single line command rather
> > > than two lines.
> > >
> > >> JB
> > >>
> > >
> > > Mike
> >
> > -------------------------------------------------------
> > SF email is sponsored by - The IT Product Guide
> > Read honest & candid reviews on hundreds of IT Products from real
users.
> > Discover which products truly live up to the hype. Start reading
now.
> > http://ads.osdn.com/?ad_id=3D6595&alloc_id=3D14396&op=3Dclick
> > _______________________________________________
> > Ccnet-devel mailing list
> > Ccn...@li...
> > https://lists.sourceforge.net/lists/listinfo/ccnet-devel
> > ******************************************************************
> > This email and any files transmitted with it from the ElPaso
> > Corporation are confidential and intended solely for the
> > use of the individual or entity to whom they are addressed.
> > If you have received this email in error please notify the
> > sender.
> > ******************************************************************
> >
> >
> >
>=20
> --
> Owen Rogers | http://dotnetjunkies.com/weblog/exortech |
> CruiseControl.NET - http://ccnet.thoughtworks.com
>=20


--=20
Owen Rogers | http://dotnetjunkies.com/weblog/exortech |
CruiseControl.NET - http://ccnet.thoughtworks.com


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id396&op=3Dick
_______________________________________________
Ccnet-devel mailing list
Ccn...@li...
https://lists.sourceforge.net/lists/listinfo/ccnet-devel
******************************************************************
This email and any files transmitted with it from the ElPaso
Corporation are confidential and intended solely for the
use of the individual or entity to whom they are addressed.
If you have received this email in error please notify the
sender.
******************************************************************

I am running CC.NET 0.9 with StarTeam 2005 as my source control.  =
Running CC.NET from the command line works great.  Running CC.NET as a =
service with my desktop user works great. =20

I have set up a 'service account', which can be used by all developers =
and managed by our network operations team. This set up does not work.  =
The account has access to the network and 'run as service'.  When CC.NET =
checks for modifications and kicks off the StarTeam command (stcmd.exe), =
the process can not find the java virtual machine 'Sun1.4.2_01'.

Questions:
Could this be starteam reqistery key issue?
Is there some other permission that needs to be granted?
Is anyone else using, a service account, cc.net, and starteam?


Sonny Rivera
son...@co...

Only CVS, although you can download zips of the latest source from CCNetLiv=
e:

http://ccnetlive.thoughtworks.com/CCNet-builds/

Mike

On 4/25/05, Marcin Hoppe <mar...@gm...> wrote:
> Hi!
>=20
> Can I fetch latest versions of CruiseControl.NET code with Subversion?
> Or is it available only via SourceForge CVS?
>=20
> It seems to me that I've seen some info on this topic on Confluence
> some time ago but I cannot find them now.
>=20
> Sorry if it is me who is blind :).
>=20
> Best regards!
> Marcin
>=20
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_ide95&alloc_id=14396&opclick
> _______________________________________________
> Ccnet-devel mailing list
> Ccn...@li...
> https://lists.sourceforge.net/lists/listinfo/ccnet-devel
>=20


--=20
mike roberts | http://mikeroberts.thoughtworks.net/ |
http://www.thoughtworks.com/

Owen,

It's the one about using Subversion revision numbers instead of dates. I
saw an email from someone (Martin???) saying that he had done something
towards it, but looking at the code I couldn't easily guess what he had
in mind. Since we're in the process of switching over from VSS to
Subversion (which is undoubtedly a step in the right direction) it would
be nice if CruiseControl played properly with it! :-)

Regards,
Richard

P.S. If the suggestion from Ruben and myself about adding extra
trigger-specific tasks is acceptable, I'd be happy to (at least try to)
implement it... Although it may take a couple of weeks.

P.P.S. I've been meaning to mention this for some time. I think there's
a typo on
http://confluence.public.thoughtworks.org/display/CCNET/Project+Configur
ation+Block - Surely it should be "artifactDirectory" not "
artifcactDirectory"?

-----Original Message-----
From: ccn...@li...
[mailto:ccn...@li...] On Behalf Of Owen
Rogers
Sent: Sunday, April 24, 2005 09:48
To: ccn...@li...
Subject: Re: [Ccnet-devel] Any progress on jira issue CCNET-398?

hi richard,
unfortunately the jira site seems to be down right now, so i'm not too
sure which jira this is.  i suspect, however, no work has been done on
it.  who was going to work on it?
cheers,
owen.

On 4/22/05, Foster, Richard - PAL <RF...@qu...> wrote:
> =20
>=20
> Greetings all,
>=20
> Subject line says it all really. Has there been any progress? If not,=20
> does anyone have any pointers about what needs to be modified and how
to test it?
>=20
> Regards,
> Richard


--
Owen Rogers | http://dotnetjunkies.com/weblog/exortech |
CruiseControl.NET - http://ccnet.thoughtworks.com


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id=14396&op=3Dick
_______________________________________________
Ccnet-devel mailing list
Ccn...@li...
https://lists.sourceforge.net/lists/listinfo/ccnet-devel

Allright, here we go then :).

This patch makes CC correctly log the perforce changenumber to the
<modifications> section. Previously it was always 0 (zero). This is
based on the 0.9 distributed version. Diffs are generated by perforce,
so I hope they are readable by whatever patch tool you guys use.

cheers
/axl

=3D=3D=3D code.diff =3D=3D=3D

---
//dice/TnT/Support/CruiseControl.NET/project/core/sourcecontrol/Perforce
/P4HistoryParser.cs	2005/04/21 13:47:47
+++
//dice/TnT/Support/CruiseControl.NET/project/core/sourcecontrol/Perforce
/P4HistoryParser.cs	2005/04/25 11:21:00
@@ -9,7 +9,7 @@
 	public class P4HistoryParser : IHistoryParser
 	{
 		private static Regex modRegex =3D new Regex(@"info1:
(?<folder>//.*)/(?<file>.*)#\d+ (?<type>\w+)");
-		private static Regex changeRegex =3D new Regex(@"text:
Change \w+ by (?<email>(?<user>.*)@.*) on (?<date>\d{4}/\d{2}/\d{2}
\d{2}:\d{2}:\d{2})");
+		private static Regex changeRegex =3D new Regex(@"text:
Change (?<change>\w+) by (?<email>(?<user>.*)@.*) on
(?<date>\d{4}/\d{2}/\d{2} \d{2}:\d{2}:\d{2})");
=20
 		/// <summary>
 		/// Used to extract changelist numbers from p4.exe
output of format
@@ -61,6 +61,7 @@
 			ArrayList mods =3D new ArrayList();
 			string line;
 			string email =3D null, user =3D null, comment =3D
string.Empty;
+			int change =3D 0;
 			DateTime date =3D DateTime.Now;
 			while((line =3D reader.ReadLine()) !=3D null)
 			{
@@ -72,6 +73,7 @@
 					mod.FolderName =3D
modificationMatch.Groups["folder"].Value;
 					mod.FileName =3D
modificationMatch.Groups["file"].Value;
 					mod.Type =3D
modificationMatch.Groups["type"].Value;
+					mod.ChangeNumber =3D change;
 					mod.EmailAddress =3D email;
 					mod.UserName =3D user;
 					mod.ModifiedTime =3D date;
@@ -84,6 +86,7 @@
 					if (changeMatch.Success)
 					{
 						// set these values
while they're available
+						change =3D
int.Parse(changeMatch.Groups["change"].Value);
 						email =3D
changeMatch.Groups["email"].Value;
 						user =3D
changeMatch.Groups["user"].Value;
 						date =3D
DateTime.Parse(changeMatch.Groups["date"].Value);


=3D=3D=3D test.diff =3D=3D=3D
=20
---
//dice/TnT/Support/CruiseControl.NET/project/UnitTests/Core/SourceContro
l/Perforce/P4HistoryParserTest.cs	2005/04/21 13:47:47
+++
//dice/TnT/Support/CruiseControl.NET/project/UnitTests/Core/SourceContro
l/Perforce/P4HistoryParserTest.cs	2005/04/25 12:13:25
@@ -16,13 +16,13 @@
 		{
 			Modification[] mods =3D
_parser.Parse(P4Mother.ContentReader, P4Mother.OLDEST_ENTRY,
P4Mother.NEWEST_ENTRY);
 			Assert.AreEqual(7, mods.Length);
-			assertModification(mods[0], "SpoonCrusher.cs",
"//depot/myproject/something", new DateTime(2002, 10, 31, 18, 20, 59),
"edit", "someone@somewhere", "someone", "Something important\r\nso
there!");
-			assertModification(mods[1], "AppleEater.cs",
"//depot/myproject/something", new DateTime(2002, 10, 31, 18, 20, 59),
"add", "someone@somewhere", "someone", "Something important\r\nso
there!");
-			assertModification(mods[2], "MonkeyToucher.cs",
"//depot/myproject/something", new DateTime(2002, 10, 31, 18, 20, 59),
"edit", "someone@somewhere", "someone", "Something important\r\nso
there!");
-			assertModification(mods[3], "IPerson.cs",
"//depot/myproject/foo", new DateTime(2002, 10, 31, 14, 20, 59), "edit",
"someone@somewhere", "someone", "One line\r\n\r\nAnother line");
-			assertModification(mods[4], "MiniMike.cs",
"//depot/myproject/foo", new DateTime(2002, 10, 31, 14, 20, 59), "add",
"someone@somewhere", "someone", "One line\r\n\r\nAnother line");
-			assertModification(mods[5], "JoeJoeJoe.cs",
"//depot/myproject/foo", new DateTime(2002, 10, 31, 14, 20, 59), "add",
"someone@somewhere", "someone", "One line\r\n\r\nAnother line");
-			assertModification(mods[6], "Fish.cs",
"//depot/myproject/tank", new DateTime(2002, 10, 31, 11, 20, 59), "add",
"bob@nowhere", "bob", "thingy\r\n(evil below)\r\nAffected files
...\r\nChange 123 by someone@somewhere on 2002/10/31 11:20:59\r\n(end of
evil)");
+			assertModification(mods[0], "SpoonCrusher.cs",
"//depot/myproject/something", new DateTime(2002, 10, 31, 18, 20, 59),
"edit", "someone@somewhere", "someone", "Something important\r\nso
there!", 3328);
+			assertModification(mods[1], "AppleEater.cs",
"//depot/myproject/something", new DateTime(2002, 10, 31, 18, 20, 59),
"add", "someone@somewhere", "someone", "Something important\r\nso
there!", 3328);
+			assertModification(mods[2], "MonkeyToucher.cs",
"//depot/myproject/something", new DateTime(2002, 10, 31, 18, 20, 59),
"edit", "someone@somewhere", "someone", "Something important\r\nso
there!", 3328);
+			assertModification(mods[3], "IPerson.cs",
"//depot/myproject/foo", new DateTime(2002, 10, 31, 14, 20, 59), "edit",
"someone@somewhere", "someone", "One line\r\n\r\nAnother line", 3327);
+			assertModification(mods[4], "MiniMike.cs",
"//depot/myproject/foo", new DateTime(2002, 10, 31, 14, 20, 59), "add",
"someone@somewhere", "someone", "One line\r\n\r\nAnother line", 3327);
+			assertModification(mods[5], "JoeJoeJoe.cs",
"//depot/myproject/foo", new DateTime(2002, 10, 31, 14, 20, 59), "add",
"someone@somewhere", "someone", "One line\r\n\r\nAnother line", 3327);
+			assertModification(mods[6], "Fish.cs",
"//depot/myproject/tank", new DateTime(2002, 10, 31, 11, 20, 59), "add",
"bob@nowhere", "bob", "thingy\r\n(evil below)\r\nAffected files
...\r\nChange 123 by someone@somewhere on 2002/10/31 11:20:59\r\n(end of
evil)", 123);
 			Assert.AreEqual(7, mods.Length);
 		}
=20
@@ -49,10 +49,11 @@
 			Assert.AreEqual("//shipping",
mods[0].FolderName);
 			Assert.AreEqual("edit", mods[0].Type);
 			Assert.AreEqual("a test comment",
mods[0].Comment);
+			Assert.AreEqual(23680, mods[0].ChangeNumber);
 		}
=20
 		private void assertModification(Modification mod, string
file, string folder,=20
-			DateTime modifiedTime, string type, string
email, string username, string comment)
+			DateTime modifiedTime, string type, string
email, string username, string comment, int change)
 		{
 			Assert.AreEqual(file, mod.FileName);
 			Assert.AreEqual(folder, mod.FolderName);
@@ -61,6 +62,7 @@
 			Assert.AreEqual(email, mod.EmailAddress);
 			Assert.AreEqual(username, mod.UserName);
 			Assert.AreEqual(comment, mod.Comment);
+			Assert.AreEqual(change, mod.ChangeNumber);
 		}
=20
 		[Test]


> -----Original Message-----
> From: ccn...@li...=20
> [mailto:ccn...@li...] On Behalf Of=20
> Marcin Hoppe
> Sent: den 25 april 2005 12:22
> To: ccn...@li...
> Subject: Re: [Ccnet-devel] Where to post patches?
>=20
> On 4/25/05, Andreas Axelsson <And...@di...> wrote:
> > I've put together a tiny patch to make the Perforce=20
> sourcecontrol plug=20
> > log the right change number. Where do I submit the patch?=20
> There's no=20
> > patch section on SourceForge.
>=20
> Send them to devel list with a short description of what does=20
> the patch actually introduce. You can also contact any of the=20
> developers, I suppose.
>=20
> Best regards!
> Marcin
>=20
>=20
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide Read honest &=20
> candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_ide95&alloc_id=14396&op=3Dick
> _______________________________________________
> Ccnet-devel mailing list
> Ccn...@li...
> https://lists.sourceforge.net/lists/listinfo/ccnet-devel
>=20

On 4/25/05, Andreas Axelsson <And...@di...> wrote:
> I've put together a tiny patch to make the Perforce sourcecontrol plug
> log the right change number. Where do I submit the patch? There's no
> patch section on SourceForge.

Send them to devel list with a short description of what does the
patch actually introduce. You can also contact any of the developers,
I suppose.

Best regards!
Marcin