As per the discussion in https://gerrit.libreoffice.org/c/core/+/89262 I would like to propose that converting Perl scripts used in Libreoffice can be ported to Python without needing any special rationale. The reasons for this are: - more people know Python than Perl - new developers (i.e. anyone under 35) do not know Perl at all - Python is a lot more readable in any case - Python has builtin functionality for a lot things where Perl needs to call external executables (like zip, tar), reducing dependencies - Converting all Perl code to Python is (in practice) a requirement for removing Cygwin as a build dependency on Windows In practice this would mean that merge requests that convert Perl to Python could just have this in their commit messages: See tdf#XXXXX for motivation.
I agree. Even if only as a pre-requisite to drop Cygwin dependency, this is already valid.
Jussi Pakkanen committed a patch related to this issue. It has been pushed to "master": https://git.libreoffice.org/core/commit/6ec05ef6fef789f5646b773495285e845e155d77 tdf#130911: convert some token generation scripts from Perl to Python. It will be available in 7.0.0. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback.
Jussi Pakkanen committed a patch related to this issue. It has been pushed to "master": https://git.libreoffice.org/core/commit/b29fb89d0a49fde0f5757774d64a7aba8298ac75 tdf#130911: convert image-sort from Perl to Python It will be available in 7.0.0. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback.
Making an easy hack out of this. Just pick one of the perl files: https://opengrok.libreoffice.org/search?project=core&type=perl
Jussi Pakkanen committed a patch related to this issue. It has been pushed to "master": https://git.libreoffice.org/core/commit/d2c23609083d7b3e5267b1e4c923476cbc509d00 tdf#130911: convert desktop-translate from Perl to Python. It will be available in 7.0.0. The patch should be included in the daily builds available at https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More information about daily builds can be found at: https://wiki.documentfoundation.org/Testing_Daily_Builds Affected users are encouraged to test the fix and report feedback.
Build fails. Linux Kubuntu, latest release. autogen.input: --with-lang=ALL --with-help=online --with-omindex=server --without-krb5 --disable-postgresql-sdbc --with-jdk-home=/usr/lib/jvm/default-java --enable-kf5 --enable-qt5 --enable-gtk3-kde5 --with-myspell-dicts Crash at Traceback (most recent call last): File "/home/tdf/git/core/solenv/bin/desktop-translate.py", line 82, in <module> assert(value[-1] == '"') AssertionError Traceback (most recent call last): File "/home/tdf/git/core/solenv/bin/desktop-translate.py", line 82, in <module> assert(value[-1] == '"') AssertionError make[1]: *** [/home/tdf/git/core/sysui/CustomTarget_share.mk:245: /home/tdf/git/core/workdir/CustomTarget/sysui/share/libreofficedev/build.flag] Error 1
Can you test if this fixes it for you: https://gerrit.libreoffice.org/c/core/+/92050
(In reply to Jussi Pakkanen from comment #7) > Can you test if this fixes it for you: > > https://gerrit.libreoffice.org/c/core/+/92050 Build in now OK with the patch
(In reply to Olivier Hallot from comment #8) > (In reply to Jussi Pakkanen from comment #7) > > Can you test if this fixes it for you: > > > > https://gerrit.libreoffice.org/c/core/+/92050 > > Build in now OK with the patch but see <https://gerrit.libreoffice.org/c/core/+/92838> "Properly escape desktop file string values"
(In reply to Jussi Pakkanen from comment #0) > As per the discussion in https://gerrit.libreoffice.org/c/core/+/89262 > > I would like to propose that converting Perl scripts used in Libreoffice can > be ported to Python without needing any special rationale. The reasons for > this are: > > - more people know Python than Perl > - new developers (i.e. anyone under 35) do not know Perl at all > - Python is a lot more readable in any case > - Python has builtin functionality for a lot things where Perl needs to call > external executables (like zip, tar), reducing dependencies > - Converting all Perl code to Python is (in practice) a requirement for > removing Cygwin as a build dependency on Windows I am not convinced that the above benefits necessarily outweigh the costs, though, or that any actual costs of conversion are at least distributed fairly in the developer base. For example, I haven't counted the hours I had to spent on <https://git.libreoffice.org/core/+/13aebf3dc75a73b381fb8d72c9788a8ee5aa911c%5E!/> "Missing dependency", <https://git.libreoffice.org/core/+/9831d806ca7223058f74c61f4e1254c84464014d%5E!/> "Fix desktop-translate.py", <https://git.libreoffice.org/core/+/009db597027890ade07d31674c740ae0fc432f87%5E!/> "Only process with brand.pl the *.keys and *.desktop files", and <https://git.libreoffice.org/core/+/3f38a57a4ed8515dd145aea46ed697f101409c2f%5E!/> "Properly escape desktop file string values" after the change from comment 5. (Though to be fair, the last two of those commits helped to find and address pre-existing issues.)
(In reply to Stephan Bergmann from comment #10) > (In reply to Jussi Pakkanen from comment #0) > > As per the discussion in https://gerrit.libreoffice.org/c/core/+/89262 > > > > I would like to propose that converting Perl scripts used in Libreoffice can > > be ported to Python without needing any special rationale. The reasons for > > this are: > > > > - more people know Python than Perl > > - new developers (i.e. anyone under 35) do not know Perl at all > > - Python is a lot more readable in any case > > - Python has builtin functionality for a lot things where Perl needs to call > > external executables (like zip, tar), reducing dependencies > > - Converting all Perl code to Python is (in practice) a requirement for > > removing Cygwin as a build dependency on Windows > > I am not convinced that the above benefits necessarily outweigh the costs, > though, or that any actual costs of conversion are at least distributed > fairly in the developer base. For example, I haven't counted the hours I > had to spent on > <https://git.libreoffice.org/core/+/ > 13aebf3dc75a73b381fb8d72c9788a8ee5aa911c%5E!/> "Missing dependency", > <https://git.libreoffice.org/core/+/ > 9831d806ca7223058f74c61f4e1254c84464014d%5E!/> "Fix desktop-translate.py", > <https://git.libreoffice.org/core/+/ > 009db597027890ade07d31674c740ae0fc432f87%5E!/> "Only process with brand.pl > the *.keys and *.desktop files", and > <https://git.libreoffice.org/core/+/ > 3f38a57a4ed8515dd145aea46ed697f101409c2f%5E!/> "Properly escape desktop file > string values" after the change from comment 5. (Though to be fair, the > last two of those commits helped to find and address pre-existing issues.) I'd like to continue converting some of the perl scripts to python. From this comment it feels like there needs to be more discussion. Am I able to start working on this easy hack? Asking about it on the dev IRC it was suggested I give a heads up before working on this.
(In reply to Jason Carty from comment #11) > I'd like to continue converting some of the perl scripts to python. From > this comment it feels like there needs to be more discussion. Am I able to > start working on this easy hack? > > Asking about it on the dev IRC it was suggested I give a heads up before > working on this. Samuel turned this "drive-by bug report" into an easy hack in comment 4. I would thus hope that he weighs in on the conversation here. (All I can personally say is what I already said in comment 10, though that is clearly unhelpful for anybody wanting to pick up this easy hack.)
Ok, I removed the easy hack from this task because of the controversy. I mostly agree with Jussi in comment 0. However, we don't want to cause too much work for other contributors. I would suggest that anyone wanting to work on this, needs to carefully select a script, which he can run and test successfully (maybe start with the bin/ folder), and also commit to handling possible regressions. Then hopefully we can get the benefits from comment 0 without too much hassle.