threads / discuss / 22701

Re: Git svn fetches entire trunk on tags (change from 1.6.5.2 to 1.7.0)

Subject: Re: Git svn fetches entire trunk on tags (change from 1.6.5.2 to 1.7.0)

## tl;dr

3 messages between Feb 17, 2010 and Feb 19, 2010.

replies: 2people: 2as markdown or json

Robert Zeh· Feb 17, 2010, 23:54 UTC · lore

On Feb 17, 2010, at 5:23 PM, Eric Wong <normalperson@yhbt.net> wrote:

> There was a rather large amount of changes between 1.6.5.2 so some
> regressions could've slipped in.  A bisection would definitely help
> us track down the cause.

I will attempt a bisection Thursday.

>
> Does the repository you're using svn:mergeinfo use by any chance?

The repository does use mergeinfo, but much later in the revision history then when the problem shows up.

Do you have any advice on how to write a test for this behaviour, since the problem is not apparent from looking at the repo? I do not like the idea of a test looking at a log but that is all I can think of right now.

Robert

Eric Wong· Feb 18, 2010, 00:06 UTC · re: Robert Zeh · lore

Robert Zeh <robert.allan.zeh@gmail.com> wrote:

Show 6 quoted lines

> On Feb 17, 2010, at 5:23 PM, Eric Wong <normalperson@yhbt.net> wrote:
>> There was a rather large amount of changes between 1.6.5.2 so some
>> regressions could've slipped in.  A bisection would definitely help
>> us track down the cause.
>
> I will attempt a bisection Thursday.

Awesome, thanks.

Show 8 quoted lines

>> Does the repository you're using svn:mergeinfo use by any chance?
>
> The repository does use mergeinfo, but    much later in the revision  
> history then when the problem shows up.
>
> Do you have any advice on how to write a test for this behaviour, since 
> the problem is not apparent from looking at the repo?  I do not like the 
> idea of a test looking at a log but that is all I can think of right now.

Andrew and Sam are the people to ask about mergeinfo, but I've found reading logs valuable for tracking down many issues in the past.

I'm not 100% convinced mergeinfo itself is the problem (but it's the code I'm least familiar with), and there were also some changes with globbing (which I believe were harmless improvements).

-- 
Eric Wong

Robert Zeh· Feb 19, 2010, 12:14 UTC · re: Eric Wong · lore

On Feb 17, 2010, at 6:06 PM, Eric Wong <normalperson@yhbt.net> wrote:

Show 9 quoted lines

> Robert Zeh <robert.allan.zeh@gmail.com> wrote:
>> On Feb 17, 2010, at 5:23 PM, Eric Wong <normalperson@yhbt.net> wrote:
>>> There was a rather large amount of changes between 1.6.5.2 so some
>>> regressions could've slipped in.  A bisection would definitely help
>>> us track down the cause.
>>
>> I will attempt a bisection Thursday.
>
> Awesome, thanks.

I now believe the real problem is that I had an ignore-paths set in the repo that fetched the entire trunk on branch. Is this expected?

Robert