You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Yasuhito FUTATSUKI <fu...@yf.bsdclub.org> on 2023/07/18 19:20:38 UTC

Re: "copy.c' line 1187: assertion failed (! svn_path_is_empty(path))" when running `svn move `

Hi,

On 2023/06/25 19:22, Daniel Sahlberg wrote:
> Den ons 31 maj 2023 kl 05:29 skrev Yasuhito FUTATSUKI <
> futatuki@yf.bsdclub.org>:
> 
>> Hello,
>>
>> On 2023/05/30 17:04, Daniel Sahlberg wrote:
>>   > Den tis 30 maj 2023 kl 04:46 skrev Jun Omae <ju...@gmail.com>:
>>
> 
> [...]
> 
> 
>>>> However, in attached patch, correct the escape sequences because "make
>>   >> check" requires at least Python 3.0.
>>
>> In the discussion in this list at 2019, there is no objection that
>> we don't support Python 3.x where x < 5, both in trunk and 1.14.x.
>> And since r1899944, we drop Python 2.7 support for make check, in trunk.
>>
>> So I think it can be accectable, at least trunk.
>>
>> However we've not drop Python 2.7 support agressively, using 'br' prefix
>> instead of 'rb' is prefered if we want back porting this to 1.14.x.
>> Because 'br' prefix can be also used in Python 2.7.
>>
>>
> I don't think this change was ever committed. Can we come to a conclusion
> on how to do it and get it committed?

I'm very sorry for replying too much late, but I don't think my post has
been blocking to commit any of them.  Any of three looks good to me.
Let's go ahead.

> In my opinion this change doesn't need to be backported to 1.14 so we can
> go with a Python 3-only solution. My reasoning is that it "only" fixes a
> SyntaxWarning so it isn't a very significant issue. By the time Python
> changes to a SyntaxError, I hope 1.14 is EOL anyway.

Even if it won't be backported to 1.14.x, there is no reason to avoid using
br'...', but also there is no strong reason to use it.

A small reasons are that we've already used br'...' in
subversion/tests/cmdline/svntest/tree.py, and I believe that using raw prefix
in regular expression string rather than using backslash escape is good for
readability.

Cheers,
-- 
Yasuhito FUTATSUKI <fu...@yf.bsdclub.org>

Re: Invalid escape sequences in Python scripts (was Re: "copy.c' line 1187: assertion failed (! svn_path_is_empty(path))" when running `svn move `)

Posted by Daniel Sahlberg <ds...@apache.org>.
On 2023/07/20 04:32:43 Yasuhito FUTATSUKI wrote:
> Jun, could you please commit the patch?

I have committed this patch now, r1912632, crediting Jun.

Thanks Jun for the patch and thanks Yasuhito for review.

Kind regards,
Daniel Sahlberg

Invalid escape sequences in Python scripts (was Re: "copy.c' line 1187: assertion failed (! svn_path_is_empty(path))" when running `svn move `)

Posted by Yasuhito FUTATSUKI <fu...@poem.co.jp>.
Hello,

On 2023/07/19 6:16, Daniel Sahlberg wrote:
> Den tis 18 juli 2023 kl 21:23 skrev Yasuhito FUTATSUKI <
> futatuki@yf.bsdclub.org>:

> I'm not fluent in Python to the level of these changes (although as far as
> I've been able to test they are good). Could one of you commit?

I looked Jun's patch in <54...@gmail.com>
and it looks good as it uses b'...' with valid escape sequence for bytes.

I also checked by using flake8 that after aplying the patch, there is
no invalid sequence warning in the scripts in subversion/ and build/
directories.

[[[
$ find . \( -name .svn -or -name __pycache__ \) -prune -or -name '*.py' -print0 | xargs -0 flake8 --select W605 | sed -e 's/:[0-9][0-9]*: W605 .*/:/' | uniq
./tools/dist/security/parser.py:58: SyntaxWarning: 'tuple' object is not callable; perhaps you missed a comma?
   (CULPRIT_SERVER,)
./contrib/client-side/svn-merge-vendor.py:429:
./contrib/client-side/svn_apply_autoprops.py:66:
./contrib/client-side/svn_apply_autoprops.py:67:
./tools/dev/analyze-svnlogs.py:38:
./tools/dev/analyze-svnlogs.py:39:
./tools/dev/analyze-svnlogs.py:40:
./tools/dev/check-license.py:26:
./tools/dev/check-license.py:27:
./tools/dev/check-license.py:28:
./tools/dev/check-license.py:29:
./tools/dev/check-license.py:30:
./tools/dev/check-license.py:31:
./tools/dev/check-license.py:32:
./tools/dev/check-license.py:33:
./tools/dev/check-license.py:34:
./tools/dev/check-license.py:35:
./tools/dev/check-license.py:36:
./tools/dev/check-license.py:37:
./tools/dev/check-license.py:38:
./tools/dev/check-license.py:39:
./tools/dev/check-license.py:40:
./tools/dev/check-license.py:41:
./tools/dev/check-license.py:42:
./tools/dev/check-license.py:43:
./tools/dev/contribulyze.py:520:
./tools/dev/contribulyze.py:523:
./tools/dev/contribulyze.py:528:
./tools/dev/contribulyze.py:579:
./tools/dev/contribulyze.py:706:
./tools/dev/mklog.py:32:
./tools/dev/po-merge.py:25:
./tools/dev/trails.py:35:
./tools/dev/trails.py:36:
./tools/po/l10n-report.py:182:
./tools/server-side/svn-backup-dumps.py:455:
]]]

I apologize if I didn't post an extra comment about it, it woluld be
commited soon.

Jun, could you please commit the patch?

Thanks,
-- 
Yasuhito FUTATSUKI <fu...@poem.co.jp>

Re: "copy.c' line 1187: assertion failed (! svn_path_is_empty(path))" when running `svn move `

Posted by Daniel Sahlberg <da...@gmail.com>.
Den tis 18 juli 2023 kl 21:23 skrev Yasuhito FUTATSUKI <
futatuki@yf.bsdclub.org>:

> Hi,
>
> On 2023/06/25 19:22, Daniel Sahlberg wrote:
> > Den ons 31 maj 2023 kl 05:29 skrev Yasuhito FUTATSUKI <
> > futatuki@yf.bsdclub.org>:
> >
> >> Hello,
> >>
> >> On 2023/05/30 17:04, Daniel Sahlberg wrote:
> >>   > Den tis 30 maj 2023 kl 04:46 skrev Jun Omae <ju...@gmail.com>:
> >>
> >
> > [...]
> >
> >
> >>>> However, in attached patch, correct the escape sequences because "make
> >>   >> check" requires at least Python 3.0.
> >>
> >> In the discussion in this list at 2019, there is no objection that
> >> we don't support Python 3.x where x < 5, both in trunk and 1.14.x.
> >> And since r1899944, we drop Python 2.7 support for make check, in trunk.
> >>
> >> So I think it can be accectable, at least trunk.
> >>
> >> However we've not drop Python 2.7 support agressively, using 'br' prefix
> >> instead of 'rb' is prefered if we want back porting this to 1.14.x.
> >> Because 'br' prefix can be also used in Python 2.7.
> >>
> >>
> > I don't think this change was ever committed. Can we come to a conclusion
> > on how to do it and get it committed?
>
> I'm very sorry for replying too much late, but I don't think my post has
> been blocking to commit any of them.  Any of three looks good to me.
> Let's go ahead.
>
> > In my opinion this change doesn't need to be backported to 1.14 so we can
> > go with a Python 3-only solution. My reasoning is that it "only" fixes a
> > SyntaxWarning so it isn't a very significant issue. By the time Python
> > changes to a SyntaxError, I hope 1.14 is EOL anyway.
>
> Even if it won't be backported to 1.14.x, there is no reason to avoid using
> br'...', but also there is no strong reason to use it.
>
> A small reasons are that we've already used br'...' in
> subversion/tests/cmdline/svntest/tree.py, and I believe that using raw
> prefix
> in regular expression string rather than using backslash escape is good for
> readability.


I'm not fluent in Python to the level of these changes (although as far as
I've been able to test they are good). Could one of you commit?

Kind regards,
Daniel