Re: [TSVN] Another go at a patch for SVNFolderStatus

At 18:54 28/10/2004 +0200, you wrote:

>I'm currently debugging the whole thing here. And as I thought it isn't
>that easy. Just checking for folder and then fetching the status
>differently (or just returning &invalidstatus) doesn't work.

Let's not do that then, let's do what it asks for.

There is definitely a bug in there are the moment, which is the following:

The OS asks for

\will\tsvntest\top\sub1
(which is a folder)

TSVN then mistakes that for a file (by falling out of the if(isFolder)
clause, chops the end off and
asks SVN for the status of the folder:

\will\tsvntest\top

Ignoring anything to do with SVN bugs, surely this wasn't what you intended?

>Well it might for the overlays alone, but there's also the column provider
>which has to use the same cache.

Right, I certainly didn't look at that possibility.

>The problem you're seeing (fetching the status from a folder above the one
>you're actually looking at) comes from the column provider which calls the
>status function on the parent folder(!) too.

What I'm seeing is that *nobody* asks for the status of the parent folder,
but TSVN mistakenly chops down the path and asks SVN for it. I'm quite
prepared to believe that the column provider does something different, and
that I've broken it, but I know what's going on in the tests I'm running here.

> The problem is that sometimes it asks the parent folder first, but
> sometimes not. So I can't just check if the status is requested by the
> column provider the first time and then do something else - sometimes the
> first call of the column provider is for a 'normal' file/folder and not
> the parent folder.

I think if we just looked up the item we were asked for that would be a
good start!

>About the relative paths: I've asked on the subversion mailing list for
>including revision 11439 in 1.1.2. Maybe you guys should kick in there and
>'vote' for that too! With that fix, we won't have to use relative paths in
>the shell extension at all, and we won't risk crashes because of that.

That would be good. But I think it's a separate issue to the
mistaking-a-path-for-a-file, which I think is going on too.

>btw: thanks for the patch. I just can't really use it.

No probs. I didn't expect you to use it as-is, it's just an unambiguous
way of talking about code changes.

I recognise that I'm made a serious communication error in combining stuff
to do with what I think is a TSVN bug, and stuff which is a work-around for
a SVN problem all into one patch.

Cheers,

Will

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tortoisesvn.tigris.org
For additional commands, e-mail: dev-help@tortoisesvn.tigris.org
Received on Thu Oct 28 20:13:42 2004