Will Dean wrote:
> "Folker Schamel" <schamel23@spinor.com> wrote in message
> news:418B8F54.7040300@spinor.com...
>
>
>>I don't see how fixing the crash itself as suggested by Steve or Julian
>>does have bad consequences for the long term.
>
>
> The argument is that it makes the repro scenario harder to identify in the
> future.
>
>
>>You still can - and have to! - find the real underlaying bug.
>
>
> Yes, but it might well be harder to do so.
-> see my answer to Karl.
>
>>You want this Windows OS crash to continue happening
>>as motivation / reminder for finding the underlaying bug?
>>
>>My impression is that Unix/Linux users are not aware
>>what a Windows Explorer crash means.
>
>
> My impression is that the implications of an access violation in a library a
> shell extension uses are being overstated.
>
> All the NT family of OS's recover fine from crashing Explorer.
Depending on your explorer preferences, a crash of the explorer
crashes all explorer windows, which alone is ugly.
More serious, windows sometimes, but not always can recover
and automatically starts the explorer again.
Admitted, I didn't have this hehaviour caused by tsvn yet
(at least as far as I know - currently I have overlways disabled
anyway because it often causes the explorer hang for several seconds),
but I have quite often that the explorer crashes in that way.
And thats REALLY ugly.
Because of this I think crashing shell extensions are a major issue.
> Some versions of Explorer have wrapped calls to shell extensions in
> structured exception handlers. There is no reason why TSVN can't make SVN
> calls from inside an SEH handler and deal with exceptions in that way.
>
> And yes, I do know something about how TSVN works...
Cheers,
Folker
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Nov 5 19:00:41 2004