Re: Build environment for official win32 builds

2012-11-20 Thread Tor Lillqvist
> yes ... but i don't remember the 4NT/native tools based build system OOo > had back in 2008 or so to be "fast" by any stretch of the imagination... Well, I have never used those, so I don't know. Surely those tools, too, tried to if not emulate, at least "simulate", POSIX in some ways? Anyway,

Re: Build environment for official win32 builds

2012-11-20 Thread Michael Stahl
On 19/11/12 19:32, Tor Lillqvist wrote: >> evidently local file system (NTFS) is rather slow, and process creation >> is very slow (which is important for any make-based build system). > > This when seen through the Cygwin POSIX emulation layer, presumably? yes ... but i don't remember the 4NT/na

Re: Build environment for official win32 builds

2012-11-19 Thread Tor Lillqvist
> Okay, but this cant explain why cygwin/gcc is so horribly slow > compared to msvc (would make both of them slow). Cygwin is a POSIX emulation. Some POSIX concepts are *very* different from Windows. Emulating them takes time and effort. > autoconf alone already is evil enough ;-) Feel free to w

Re: Build environment for official win32 builds

2012-11-19 Thread Enrico Weigelt
> evidently local file system (NTFS) is rather slow, and process > creation > is very slow (which is important for any make-based build system). Okay, but this cant explain why cygwin/gcc is so horribly slow compared to msvc (would make both of them slow). > > Does this also mean kicking out aut

Re: Build environment for official win32 builds

2012-11-19 Thread Tor Lillqvist
> evidently local file system (NTFS) is rather slow, and process creation > is very slow (which is important for any make-based build system). This when seen through the Cygwin POSIX emulation layer, presumably? And when using a build mechanism that has been developed on Unix systems in general an

Re: Build environment for official win32 builds

2012-11-19 Thread Michael Stahl
On 16/11/12 20:30, Enrico Weigelt wrote: >> The real question is - what is the deliverable here ? building on >> windows under cygwin is -horribly- slow - if we can use the same >> compiler to cross-compile from Linux (as we can) and build around 50x >> faster - why would we want to do work &

Re: Build environment for official win32 builds

2012-11-16 Thread Enrico Weigelt
Hi, > Anything is possible in LibreOffice :-) having said that - > personally > I'd want to see that adding support for this doesn't tangle the build > system too badly - particularly wrt. confusing our mingw > cross-compilation support. I have to admit, I haven't tried it yet, but _a

Re: Build environment for official win32 builds

2012-11-16 Thread Michael Meeks
On Fri, 2012-11-16 at 15:32 +0100, Enrico Weigelt wrote: > > in addition to above points also there are also open questions around > > how we would run unit tests in a MinGW cross-compilation environment > > and how well gdb works on Windows; the MSVC C++ debugger is really quite > > good. > > Wh

Re: Build environment for official win32 builds

2012-11-16 Thread Enrico Weigelt
Hi, > There are also some features in the code that don't compile with > MinGW. They use API that MinGW does not provide headers for etc. Ok. Is there already some list of known issues ? > Let me also point out that we definitely do not want to build > *locally* on Windows with GCC (MinGW). MinG

Re: Build environment for official win32 builds

2012-11-16 Thread Enrico Weigelt
> in addition to above points also there are also open questions around > how we would run unit tests in a MinGW cross-compilation environment > and how well gdb works on Windows; the MSVC C++ debugger is really quite > good. What about native compiling (not crosscompile) on win32 using gcc ? c

Re: Build environment for official win32 builds

2012-11-16 Thread Michael Stahl
On 16/11/12 12:12, Tor Lillqvist wrote: >>> Is there some specific reason for using MSVC instead of gcc ? >> >> Currently ABI backwards-compatibility for compiled extensions. > > There are also some features in the code that don't compile with > MinGW. They use API that MinGW does not provide hea

Re: Build environment for official win32 builds

2012-11-16 Thread Tor Lillqvist
>> Is there some specific reason for using MSVC instead of gcc ? > > Currently ABI backwards-compatibility for compiled extensions. There are also some features in the code that don't compile with MinGW. They use API that MinGW does not provide headers for etc. Let me also point out that we defi

Re: Build environment for official win32 builds

2012-11-16 Thread Michael Meeks
Hi Enrico, On Fri, 2012-11-16 at 11:22 +0100, Enrico Weigelt wrote: > > and use the configure options from > > distro-configs/LibreOfficeWin32.conf > > Thanks. > > Is there some specific reason for using MSVC instead of gcc ? Currently ABI backwards-compatibility for compiled extensions

Re: Build environment for official win32 builds

2012-11-16 Thread Enrico Weigelt
> See > http://wiki.documentfoundation.org/Development/Windows_Build_Dependencies > > and use the configure options from > distro-configs/LibreOfficeWin32.conf Thanks. Is there some specific reason for using MSVC instead of gcc ? Is it possible to build w/ Qt or Gtk instead of native widget set

Re: Build environment for official win32 builds

2012-11-16 Thread Miklos Vajna
On Thu, Nov 15, 2012 at 11:44:38AM +0100, Enrico Weigelt wrote: > I'd like to setup an win32 build environment which reproduces > the official win32 builds for verifying my patches. > > What do I need for that ? See http://wiki.documentfoundation.org/Development/Windows_Build_Dependencies and

Build environment for official win32 builds

2012-11-15 Thread Enrico Weigelt
Hi folks, I'd like to setup an win32 build environment which reproduces the official win32 builds for verifying my patches. What do I need for that ? thx -- Mit freundlichen Grüßen / Kind regards Enrico Weigelt VNC - Virtual Network Consult GmbH Head Of Development Pariser Platz 4a, D-10