Re: SP in committer line in fast-import stream

From: Dmitry Ivankov <divanorama@gmail.com>
Date: Jul 18, 2011, 15:38 UTC
Message-ID: <loom.20110718T172927-173@post.gmane.org>
In-Reply-To: <CAE3X6mwJquoHj06FVGTsg0qtzyTwbd6gNqy7J4yWiVF-+p-23Q@mail.gmail.com>

Hi,

SASAKI Suguru <sss.sonik <at> gmail.com> writes:

Show 41 quoted lines

> 
> Hi,
> 
> I'm working with data from `bzr fast-export` and `git fast-import`.
> (bzr is 2.4b5, git is 1.7.5.4, on Debian GNU/Linux (sid))
> 
> Export and import themselves are OK,
> but `git fsck --strict` exits with error, saying:
> 
>   error in commit 2e7a16fbe57b555c1c5954470ef66f3a2a089288: invalid
> author/committer line - missing space before email
> 
> and pushing to remote like GitHub fails.
> 
> I found minimal OK-data unlike `bzr fast-export` outputs and NG-data
> like `bzr fast-export`.
> (Attached: test_NG.data.txt and test_OK.data.txt)
> 
> Only one difference between these is a space in committer line.
>   * OK: 'committer' SP SP LT GT ...
>   * NG: 'committer' SP    LT GT ...
> 
> `man git-fast-import` says:
> 
>   commit
>     Create or update a branch with a new commit, recording one logical
> change to the project.
> 
>       'commit' SP <ref> LF
>       mark?
>       ('author' (SP <name>)? SP LT <email> GT SP <when> LF)?
>       'committer' (SP <name>)? SP LT <email> GT SP <when> LF
>       data
>       ('from' SP <committish> LF)?
>       ('merge' SP <committish> LF)?
>       (filemodify | filedelete | filecopy | filerename | filedeleteall
> | notemodify)*
>       LF?
> 
> I think, from this notations, both data is OK.
> What's the problem?

The problem is with git-fast-import that it doesn't verify the format strictly here. For example following (no LT) will pass: <name> SP <email> GT The second problem is that it generates "bad" committer, in fact name-email is used as-is, so at least it should convert absent name to a empty name. Or maybe just fix the format to make string obligatory. There even is a third minor problem, fsck will report confusing "missing space" for the no-LT example.

Third one is a clear. Your one is the second one, while internally it pulls the first one too.

The shortest fix is to read documentation as 'committer' SP <name> SP LT <email> GT SP <when> LF

> 
> Regards,
>