[PATCH 1 of 21] Thread isolation in hgweb

[Augie Fackler]

On Tue, Sep 1, 2015 at 1:15 PM, Gregory Szorc <gregory.szorc at gmail.com> wrote:
> Was anyone going to look at this or should I attempt a more traditional,
> partial patchbomb?

Something of this magnitude might actually be good sprint material, if
you can wait that long.

>
> On Sat, Aug 22, 2015 at 7:01 PM, Gregory Szorc <gregory.szorc at gmail.com>
> wrote:
>>
>> I put together a too-large-to-patchbomb series for making hgweb more
>> thread safe. Most of the patches are refactoring so all the request-specific
>> state is isolated on a single object and not shared on the hgweb / WSGI
>> application instance. The final patches establish a cached local repo class
>> that knows how to "refresh" a localrepository instance when necessary.
>> Finally, we establish a pool of repos so hgweb can use a unique
>> localrepository per thread. This was what mpm said was a prerequisite to
>> dropping purported thread safety of localrepository.
>>
>> To obtain the series:
>>
>> hg pull -r f0a9a8b5bafa https://hg.mozilla.org/users/gszorc_mozilla.com/hg
>>
>> I haven't explicitly tested to see if this prevents some of the known
>> threading bugs. I'll have to load up siege to verify. I would say most of
>> the patches are low-risk until the final 4. If something breaks: I expect it
>> to be from those 4.
>
>
>
> _______________________________________________
> Mercurial-devel mailing list
> Mercurial-devel at selenic.com
> https://selenic.com/mailman/listinfo/mercurial-devel
>