Hi Henrik,
At 22.46 01/09/2006, Henrik Nordstrom wrote:
fre 2006-09-01 klockan 19:29 +0200 skrev Guido Serassio:
> What is your opinion to also add MinGW support to 2.6 ?
It's a bit too late for that. It's a STABLE branch so no intrusive
changes unless absolutely needed to fix a bug..
OK. However the needed changes are not so intrusive, the majority of
the code is isolated in three files: include/squid_mswin.h,
lib/win32lib.c and src/ipc_win32.c plus some isolated changes to
unlinkd.c, pinger.c and logfile-daemon.c. And of course, some
isolated changes to configure.in and makefiles.
On the 3.0 side things are something different: the changes can be
more intrusive if we will use the IPC refactoring currently included
in the nt 3.0 branch.
We have two choices here:
1. Merge also the IPC refactoring in 3.0 with MinGW support
2. "rollback" the IPC refactoring work from nt 3.0 branch and forward
port the ipc_win32.c file from the nt 2.6 branch into a new ipc_win32.cc
Solution 1 is more intrusive on the IPC side, and the code is not so
well tested, but it's a definitive solution.
Solution 2 is less intrusive, and the code should be safer, but this
could be only a temporary solution waiting for 3.1.
The other changes needed for 3.0 are pretty similar to 2.6 changes,
with some change already merged.
Regards
Guido
-
========================================================
Guido Serassio
Acme Consulting S.r.l. - Microsoft Certified Partner
Via Lucia Savarino, 1 10098 - Rivoli (TO) - ITALY
Tel. : +39.011.9530135 Fax. : +39.011.9781115
Email: [EMAIL PROTECTED]
WWW: http://www.acmeconsulting.it/
