[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [TSVN] RFC: New cache scheme

From: Will Dean <svn_at_indcomp.co.uk>
Date: 2005-01-21 15:49:42 CET

At 15:41 21/01/2005 +0100, you wrote:

>Yes, it would be a security issue. So called priviledge escalations
>might be possible.

Yes, although the flip-side is that LOCALSYSTEM is actually less privileged
in terms of network access.

>The advantage of a service isn't just that it's loaded before someone
>is logged on. It's also a way to reliably (without users being able to
>interfere/disable it) start it automatically. Every other method is
>easy for a halfway experienced user to remove (autostart, registry,
>...).

I'm not sure I'm convinced. You can stop a service automatically starting
from the U/I or the registry.

But anyway, if the shell autostarts the cache if it needs it, who cares?

>But that's something we can decide later.

Yes, if somebody wants to put up with all the SCM crap, they can do that.

>I've seen that you haven't changed the column provider yet to use your
>new process. As soon as you do, you'll have to store more status info
>for each entry (e.g. the URL) and the memory use will increase
>slightly. But I don't think it will increase a lot.

It is using it, actually, because the SVNFolderStatus call it asking the
remote cache. But as you've pointed out, and I hadn't noticed, it's now
broken because of the missing entry info.

>If we ever(?) want to start the cache process as a service, then the
>UI has to be in TortoiseProc. A service never should pop up an UI for
>security reasons.

Yes (not sure I agree about the reasons, but certainly would need to be
elsewhere if we were using a service)

>Not really. It would require a lot of rewriting the existing code, but
>that's all.

Would it? We could make an ATL COM object to carry the property page in
about three mouse-clicks and then *just* (I hate it when people say that to
me) move the code across...

>Yes. But that error can also happen under other circumstances, so it's
>not 100% sure why the error happens.

But we could use it to queue a deferred update of that item.

>Looks very good so far. I'll have a closer look this weekend.

I'm currently re-writing it to change the way directory status is
stored. So don't learn it too diligently!

Cheers,

Will

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tortoisesvn.tigris.org
For additional commands, e-mail: dev-help@tortoisesvn.tigris.org
Received on Fri Jan 21 15:50:33 2005

This is an archived mail posted to the TortoiseSVN Dev mailing list.