Inconsistency in blame

[Arwin Arni <ar...@collab.net>]

Hi All,

I recently discovered what appears to be a bug.

When I blame a file that was added to a branch (say branch A) via a 
merge (from, say branch B), the revisions column is populated with the 
blame of the original file (as it is in branch B).

I would expect it to be populated with the revision in which it was 
merged to branch A.

blame -g in turn shows me the merge history of the original file (from 
branch B).



Here's a working example:

svn log 
http://svn.apache.org/repos/asf/subversion/trunk/subversion/libsvn_subr/cache-membuffer.c

This file was added to trunk only in r1067669 via a merge from 
branches/performance

However, a "blame" shows me revisions that are older than this (same as 
blame of the file in branches/performance), and not the revision in 
which the file was added.

A "blame -g" shows me the merge history of the file in branches/performance.

This has been the case since 1.4.x. (That's as far back as I have tested 
it.)

Should this be documented somewhere/ should the behavior be changed to 
be uniform for both newly added files and modified files?


Regards,
Arwin Arni

Re: Inconsistency in blame

[Arwin Arni <ar...@collab.net>]

On Monday 07 February 2011 06:37 PM, Arwin Arni wrote:
> Hi All,
>
> I recently discovered what appears to be a bug.
>
> When I blame a file that was added to a branch (say branch A) via a 
> merge (from, say branch B), the revisions column is populated with the 
> blame of the original file (as it is in branch B).
>
> I would expect it to be populated with the revision in which it was 
> merged to branch A.
>
> blame -g in turn shows me the merge history of the original file (from 
> branch B).
>
>
>
> Here's a working example:
>
> svn log 
> http://svn.apache.org/repos/asf/subversion/trunk/subversion/libsvn_subr/cache-membuffer.c 
>
>
> This file was added to trunk only in r1067669 via a merge from 
> branches/performance
>
> However, a "blame" shows me revisions that are older than this (same 
> as blame of the file in branches/performance), and not the revision in 
> which the file was added.
>
> A "blame -g" shows me the merge history of the file in 
> branches/performance.
>
Sorry, the file was merged from branches/integrate-cache-membuffer
> This has been the case since 1.4.x. (That's as far back as I have 
> tested it.)
>
> Should this be documented somewhere/ should the behavior be changed to 
> be uniform for both newly added files and modified files?
>
>
> Regards,
> Arwin Arni