Re: Make the git codebase thread-safe
On Sat, Feb 15, 2014 at 8:15 AM, Zachary Turner <zturner@chromium.org> wrote:
Show 13 quoted lines
> Even if we make that change to use TLS for this case, the
> implementation of pread() will still change in such a way that the
> semantics of pread() are different on Windows. Is that ok?
>
> Just to summarize, here's the viable approaches I've seen discussed so far:
>
> 1) Use _WINVER at compile time to select either a thread-safe or
> non-thread-safe implementation of pread. This is the easiest possible
> code change, but would necessitate 2 binary distributions of git for
> windows.
> 2) Use TLS as you suggest and have one fd per pack thread. Probably
> the most complicated code change (at least for me, being a first-time
> contributor)
It's not so complicated. I suggested a patch [1] before (surprise!).
> 3) Use Karsten's suggested implementation from earlier in the thread.
> Seems to work, but it's a little confusing from a readability
> standpoint since the implementation is not-thread safe except in this
> specific usage contex
[1] http://article.gmane.org/gmane.comp.version-control.git/196042
--
Duy
Duy Nguyen <pclouds@gmail.com>