Philip Martin wrote:
> Branko Čibej <brane@xbc.nu> writes:
>
>
>> Garrett Rooney wrote:
>>
>>> On 7/31/06, Philip Martin <philip@codematters.co.uk> wrote:
>>>
>>>
>>>> Do any developers regularly build/test using --enable-dso? Has anyone
>>>> ever done any stress testing with --enable-dso?
>>>>
>>> Personally, I've never done so, I don't know if anyone else does.
>>>
>> Since we're not actually testing that stuff, and it seems to be a huge
>> potential problem, why don't we just disable DSO in 1.4?
>>
>
> When I asked last year Greg Hudson replied that some packagers use it:
>
> http://svn.haxx.se/dev/archive-2005-07/1344.shtml
>
> and we have had a small number of bug reports from users that were a
> result of --enable-dso, although that could be a reason to disable it.
>
>
>> I think it's a
>> misfeature the way it's implemented now, anyway.
>>
>
> Given that Subversion with --disable-dso still tries to load DSOs I
> tend to agree.
>
I think that --enable-dso as it stands now is more trouble than it's
worth. It will only work on ELF systems (even Mac OS X or HP-UX won't
work, and I won't mention Windows at all), and the implementation has
several obvious bugs, pool handling being one of them. (And yes, I agree
with Justin that this is about RA/FS loaders.)
Given that, and the fact that I don't see a realistic chance of fixing
the implementation in time for 1.4.0, I vote we disable dynamic RA/FS
module loading in 1.4.0, and re-enable it only after
* the implementation is fixed
* we have at least _one_ buildbot testing this feature regularly
If we all agree, I volunteer to do the first step -- disabling dynamic
RA/FS loading -- and I'll certainly help with fixing things later on,
especially on Windows.
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Aug 1 01:17:34 2006