On 20/01/2012 10:08 p.m., Kinkie wrote:
On Fri, Jan 20, 2012 at 8:17 AM, Amos Jeffries<squ...@treenet.co.nz> wrote:
On 20/01/2012 10:20 a.m., Kinkie wrote:
My understanding is that no new/special dummy code is needed for Squid
on Windows in non-SMP mode. Bugs notwithstanding, non-SMP code should
work as well as before shared caching and other SMP features were
committed.
Thanks for the details. I guess we need an updated MinGW test run to see
where the particular issues are and see if and why they exist.
Currently trunk doesn't build on windows at all, mostly stuff at the
winsock api level.
The mswin branch
(http://bzr.squid-cache.org/bzr/squid3/branches/mswin) is currently a
2.6k lines patch, and brings things a little further ahead, but not
quite enough yet as you can see here:
http://build.squid-cache.org/job/mswin-mingw/175/console
But it doesn't seem to make things worse on other platforms, so I'd
consider merging back to trunk.
The mswin branch itself is a bit messy as we overlapped and incrementally
fixed pieces. There are still some experiments in there that got pasted over
with better fixes that made the experimental code irrelevant without
removing the experimental bits.
So I think it would be better to cherry-pick the changes to trunk and
document each better with commit messages.
I can handle that.
In most cases what is needed is extra header files. Should I go
through the normal review process or are these harmless enough to be
merged directly?
Thank you. Normal process I think for this part, just in case we miss
something in the picking :)
I have a bit of change half done to polish up PRIuSIZE, but the rest of
the branch is open pickings.
I've kicked off another 3.ALPHA-branch-matrix test to verify its working
everywhere, the last few runs failed in the purge code, so its hard to
tell if there was anything else later on.
Amos
