[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [jgit-dev] PathMatcher behaviour is inconsistent with "git check-attr"
|
Hello Matthias,
Thanks for the guide. I've created an account at accounts.eclipse.org and signed the legal stuff.
But when I try to login with the same login/password to Gerrit, it says
Invalid username or password.
I'm pretty sure I specify the same login/password pair as for accounts.eclipse.org. Isn't Gerrit
account automatically created with accounts.eclipse.org account?
--
Dmitry Pavlenko,
TMate Software,
http://subgit.com/ - git-svn bridge
On Wednesday 28 June 2017 21:30:01 Matthias Sohn wrote:
> On Wed, Jun 28, 2017 at 9:11 PM, Dmitry Pavlenko <pavlenko@xxxxxxxxxxxxx>
>
> wrote:
> > Hello,
> > There's a case when PathMatcher behaviour differs from "git check-attr".
> > If I create .gitattributes
> > file with the following content:
> >
> > images/** attr=value
> >
> > and then run
> >
> > $ git check-attr attr images
> > images/: attr: unspecified
> >
> > $ git check-attr attr images/
> > images/: attr: unspecified
> >
> > $ git check-attr attr images/x
> > images/x: attr: value
> >
> > If I do the same with PathMatcher.createPathMatcher(), with every
> > reasonable parameters it says that
> > all the paths above match to the pattern "images/**". So it's an
> > inconsistency with Git behaviour.
> >
> > It's not a surprise because near line 234 of PathMatcher.java there's a
> > comment
> >
> > if (matcher == matchers.size() - 2
> >
> > &&
> >
> > matchers.get(matcher + 1) == WILD)
> >
> > // ** can match *nothing*:
> > a/b/** match also a/b
> >
> > return true;
> >
> > stating that a/b/** should match a/b, what's wrong.
> >
> > I've attached a patch fixing the problem for all my cases, but I was
> > surprised why the code has
> >
> > matchers.size() - 2
> >
> > and
> >
> > matcher + 1
> >
> > instead of
> >
> > matchers.size() - 1
> >
> > and
> >
> > matcher
> >
> > I've changed it for the latter and the code still works. For me it makes
> > more sense as it checks
> > whether current segment is double star and it is actually the last one.
> >
> > Could you please have a look at those 4 lines and maybe apply my changes?
>
> Thanks for your bug report and patch contribution.
>
> Please sign the Eclipse contribution license agreement (CLA) [1] and follow
> the contributor guide [2]
> to push your patch to Gerrit which we use for code review.
>
> [1] https://eclipse.org/legal/CLA.php
> https://accounts.eclipse.org/user/838/eca
> [2] https://wiki.eclipse.org/EGit/Contributor_Guide#Contributing_Patches
>
> -Matthias
