Re: svn commit: r1915317 - /subversion/branches/1.14.x/STATUS
Hello, On 2024/02/02 17:45, Daniel Sahlberg wrote: > There is also the rewrite of mailer.py hook-script (in tools/hook-scripts). > I'm not sure if it is complete yet and I'm not sure if we want to release > it in a patch release. We've said before not to break Python2 > compatibility, on the other hand the current script doesn't work at all > under Python3 and I'm leaning towards Python3 is the more important target > at this moment. @Greg Stein can probably comment more on > the current status. mailer.py in 1.14.x before r1915337 was indeed broken for Python 3 since 1.14.0 was released. However after r1885829, including 1.14.1 - 1.14.3, it worked if * do not use "for_path" definition in group setting on mailer.conf. or * do not use it for revpropchange/lock/unlock notification. So it is not "at all". Actually, I've used it with Python 3.6 since 2021-03-03(at latest) but those conditions are satisfid. Also my local test did not cover opposit of the former condition. That's why it wasn't fixed since then until r1915337, after 1.14.3 was released. Anyways, I'd be glad if those fix would released with Subversion 1.14.4. Cheers, -- Yasuhito FUTATSUKI
Re: svn commit: r1915317 - /subversion/branches/1.14.x/STATUS
Den fre 2 feb. 2024 kl 08:20 skrev Jun Omae : > On Thu, Feb 1, 2024 at 10:42 PM Daniel Sahlberg > wrote: > > > > Den fre 19 jan. 2024 kl 07:41 skrev : > >> > >> Author: jun66j5 > >> Date: Fri Jan 19 06:40:59 2024 > >> New Revision: 1915317 > >> > >> URL: http://svn.apache.org/viewvc?rev=1915317&view=rev > >> Log: > >> * STATUS: Nominate r1915316. > >> > >> Modified: > >> subversion/branches/1.14.x/STATUS > >> > >> Modified: subversion/branches/1.14.x/STATUS > >> URL: > http://svn.apache.org/viewvc/subversion/branches/1.14.x/STATUS?rev=1915317&r1=1915316&r2=1915317&view=diff > >> > == > >> --- subversion/branches/1.14.x/STATUS (original) > >> +++ subversion/branches/1.14.x/STATUS Fri Jan 19 06:40:59 2024 > >> @@ -45,6 +45,14 @@ Candidate changes: > >> Votes: > >> +1: dsahlberg > >> > >> + * r1915316 > >> + swig-py: Fix `none_dealloc` error caused by reference count issue in > >> + `apply_textdelta` invoked from `svn.repos.replay()`. > >> + Justification: > >> + Fix Python interpreter crash which often occurs with large > repositories. > >> + Votes: > >> + +1: jun66j5 > >> + > >> Veto-blocked changes: > >> = > > > > > > Is this serious enough that we should consider a 1.14.4 release? (Didn't > have time to review the code yet but a quick glance positive). > > > > Kind regards, > > Daniel > > In my opinion, yes. Python users have no workaround for the issue > except downgrading to 1.14.2. I think it happens if Python is a > long-running process like a daemon or with large repositories. > > At least, I'm considering to report it with the patch to distributors > (Debian, Ubuntu, FreeBSD, etc.). > > -- > Jun Omae (大前 潤) > Alright, since Nathan also thought about a release we should probably do this. It seems the above fix is already backported (r1915338). There are two other fixes with two votes in the STATUS file: * r1914222 Improve help message for svnmucc PUT. This is only a help message change, but it is within the "core code" so I think someone else has to approve it as well (but we can probably bend the rules since it only affects translations). * r1912632 Fix `invalid escape sequence` in Python scripts to prevent many `SyntaxWarning`s since Python 3.12. This is the patch created by Jun last summer, review by futatuki and committed by me. Jun, would you consider voting for this? The two above should be fairly easy fixes and I think it would be good to include. Then we have the following group: * r1890223, r1890668, r1890673 Support building on Win64/ARM64. It is not approved and I'm not sure if we have someone who can look at it. There is also the rewrite of mailer.py hook-script (in tools/hook-scripts). I'm not sure if it is complete yet and I'm not sure if we want to release it in a patch release. We've said before not to break Python2 compatibility, on the other hand the current script doesn't work at all under Python3 and I'm leaning towards Python3 is the more important target at this moment. @Greg Stein can probably comment more on the current status. Kind regards, Daniel
Re: svn commit: r1915317 - /subversion/branches/1.14.x/STATUS
On Thu, Feb 1, 2024 at 10:42 PM Daniel Sahlberg wrote: > > Den fre 19 jan. 2024 kl 07:41 skrev : >> >> Author: jun66j5 >> Date: Fri Jan 19 06:40:59 2024 >> New Revision: 1915317 >> >> URL: http://svn.apache.org/viewvc?rev=1915317&view=rev >> Log: >> * STATUS: Nominate r1915316. >> >> Modified: >> subversion/branches/1.14.x/STATUS >> >> Modified: subversion/branches/1.14.x/STATUS >> URL: >> http://svn.apache.org/viewvc/subversion/branches/1.14.x/STATUS?rev=1915317&r1=1915316&r2=1915317&view=diff >> == >> --- subversion/branches/1.14.x/STATUS (original) >> +++ subversion/branches/1.14.x/STATUS Fri Jan 19 06:40:59 2024 >> @@ -45,6 +45,14 @@ Candidate changes: >> Votes: >> +1: dsahlberg >> >> + * r1915316 >> + swig-py: Fix `none_dealloc` error caused by reference count issue in >> + `apply_textdelta` invoked from `svn.repos.replay()`. >> + Justification: >> + Fix Python interpreter crash which often occurs with large >> repositories. >> + Votes: >> + +1: jun66j5 >> + >> Veto-blocked changes: >> = > > > Is this serious enough that we should consider a 1.14.4 release? (Didn't have > time to review the code yet but a quick glance positive). > > Kind regards, > Daniel In my opinion, yes. Python users have no workaround for the issue except downgrading to 1.14.2. I think it happens if Python is a long-running process like a daemon or with large repositories. At least, I'm considering to report it with the patch to distributors (Debian, Ubuntu, FreeBSD, etc.). -- Jun Omae (大前 潤)
Re: svn commit: r1915317 - /subversion/branches/1.14.x/STATUS
Den fre 19 jan. 2024 kl 07:41 skrev : > Author: jun66j5 > Date: Fri Jan 19 06:40:59 2024 > New Revision: 1915317 > > URL: http://svn.apache.org/viewvc?rev=1915317&view=rev > Log: > * STATUS: Nominate r1915316. > > Modified: > subversion/branches/1.14.x/STATUS > > Modified: subversion/branches/1.14.x/STATUS > URL: > http://svn.apache.org/viewvc/subversion/branches/1.14.x/STATUS?rev=1915317&r1=1915316&r2=1915317&view=diff > > == > --- subversion/branches/1.14.x/STATUS (original) > +++ subversion/branches/1.14.x/STATUS Fri Jan 19 06:40:59 2024 > @@ -45,6 +45,14 @@ Candidate changes: > Votes: > +1: dsahlberg > > + * r1915316 > + swig-py: Fix `none_dealloc` error caused by reference count issue in > + `apply_textdelta` invoked from `svn.repos.replay()`. > + Justification: > + Fix Python interpreter crash which often occurs with large > repositories. > + Votes: > + +1: jun66j5 > + > Veto-blocked changes: > = > Is this serious enough that we should consider a 1.14.4 release? (Didn't have time to review the code yet but a quick glance positive). Kind regards, Daniel
