[Pidgin] BuildingWinPidgin/3.0.0 modified
lrn1986 at gmail.com
Sat Apr 26 13:06:22 EDT 2014
-----BEGIN PGP SIGNED MESSAGE-----
On 26.04.2014 20:55, Tomasz Wasilczyk wrote:> On 04/26/2014 06:23 PM, Richard
>> On Sat, 2014-04-26 at 11:44 +0200, Tomasz Wasilczyk wrote:
>>> it should still be possible to build on Windows using cygwin *and*
>> Have you tested that? I think it'd be really frustrating to be on
>> Windows trying to do some development and have to keep cross-building
>> from another box.
> Personally - no, I don't like Cygwin. But LRN, who sent us initial patches
> for fixing autotools-based win32 buildsystem, used it.
I'm using MSYS2, not Cygwin. In a nutshell, the difference is that with
Cygwin you have to cross-compile to build a W32 binary (if you don't, you
build a Cygwin binary that relies on Cygwin runtime), whereas in MSYS2 you
don't need to cross-compile. Also, since with MSYS2 you can (and should) use
W32 tools at build time (including gcc), the building process may be a bit
faster, since W32 tools don't pay the Cygwin runtime performance cost.
> And you don't have to use two boxes. I use linux machine with Win7
> launched within virtualbox. You can use Windows machine with virtualboxed
This is actually viable. I did something like this once, and building stuff
in a VM that runs on machine X under Windows is MUCH *FASTER* than building
the same stuff on machine X under Windows with MSYS2 or Cygwin (and somewhat
faster than building the same stuff on machine X under Windows with a
buildsystem that is not autotools and does not need a compatibility layer).
Apparently, the cost of virtualization is lower than the cost of emulating
fork() and other POSIX syscalls on top of W32.
O< ascii ribbon - stop html email! - www.asciiribbon.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (MingW32)
-----END PGP SIGNATURE-----
More information about the Devel