Jeremy Nicoll

On Sun, 9 Nov 2025, at 14:17, Dom Wise wrote: Windows doesn't have any concept of an owner process for named objects like shared mutexes. ... when a process ends Windows will automatically close any handles that the process did not already close, though it won't release ownership of any mutexes still held If processes don't own mutexes, what does? Can one ask the OS who/what owns one? Or who/what created it & when? -- Jeremy Nicoll - my opinions are my own.

On Fri, 13 Jun 2025, at 19:42, m-stgt wrote: While other application languages list [editors which support it] ooRexx just mentions in 'ooRexx Documentation 5.0.0 Open Object Rexx Programmer Guide', paragraph '2.4 Writing Your Program', p. 9 Windows Notepad or Gnome gedit (on Linux) are a couple widely available editors. No hint about smart auto-indent, no downloads for syntax highlighting, minimalist text edior and that's it... I suppose if one was new to any sort of programming on a PC & ooRexx...

On Fri, 6 Jun 2025, at 18:46, m-stgt wrote: Today the failure is on the PC (x64) in clause ~~~ if RxFuncAdd('RxPipe', 'RXPIPE', 'RXPIPE') > 0, / see PIPE.HTM / So... that worked, or /seemed/ to. Does RxFuncAdd do more than find a DLL, load it into memory, & check that it has the named entry-point? If the DLL was for the wrong architecture, would RxFuncAdd fail? I know nothing about how to use (or test) RxPipe. -- Jeremy Nicoll - my opinions are my own.

On Fri, 6 Jun 2025, at 03:30, m-stgt wrote: ooRexx 5.oo and younger are availabe as portable app. A program I once developed using ooRexx V4.1.3 (and IIRC still worked with 5.oo) now fails. How may I get an ooRexx version before 5.oo set up as portable app? I'm not sure that you can, possibily because of the way the rxapi process worked back then. But, you might - if you have no permanently-installed version of ooRexx now be able to (temporarily) permanently (ie fully) install v4.1.3 & then use the...

On Tue, 25 Mar 2025, at 12:42, Aleksey Tarasow wrote: SemTran=SysCreateMutexSem('/SEM32/'translate(!global.!TrFile, '', ':.*','_')) do while SysRequestMutexSem(SemTran, 10)<>0; end TranFile=.stream~new(!global.!TrFile) Tran=TranFile~lineout (‘test’) TranFile~close call SysReleaseMutexSem SemTran; call SysCloseMutexSem SemTran ~~~ Question. Is this an optimal solution or not? Are there other options for solving the problem? It depends (maybe) on how much waiting the tasks wishing to write to the file...

On Wed, 5 Mar 2025, at 00:46, m-stgt wrote: Wouldn't that match eg "55.1234.17.9.23" which doesn't look like an ok number to me? Yes, it would. But this is (for me) a bearable flaw in my goal to allow dots and commas as decimal separators (checked later which one actually is decimal and thousands separator) and dots, commas, blanks, and apostrophe as thousands separators. Ah, ok. In your place I think I'd have the user choose a single set of conventions - it's mainly a country-specific thing is it...

On Tue, 4 Mar 2025, at 23:03, Rick McGuire wrote: Sorry, I meant to say "zero or one times". I'd be surprised. Inasmuch as there's any standard syntax for forms of RE, before PCRE etc came along, I think it was normal for to mean zero or more occurrences of what went before; to mean one or more occurrences of what went before I'm nearly certain this is what I was taught in the late 1970s /early 1980s. (But goodness knows what that was based on.) On the Regex Buddy site, there's a tool which helps...

On Tue, 4 Mar 2025, at 23:01, Rick McGuire wrote: According to the documentation, * means match the previous expression 0 or more times. I suspect -* would accomplish that. But one doesn't want to match strings like "-----------------------------7" when just wanting to match "8" or "-1924". -- Jeremy Nicoll - my opinions are my own.