Jens-Heiner Rechtien wrote:
Curtis Clauson wrote:
[EMAIL PROTECTED] wrote:
Hi Curtis,
Curtis Clauson zei:
Pavel Janík wrote:
From: Curtis Clauson <[EMAIL PROTECTED]> Date: Sun, 30 Jan 2005 03:49:04 -0800
> > Can OOo for windows be built with GNU compilers in Cygwin instead
of
> > using MS Visual Studio? It seems like it should, but all of the
build
> > docs only talk about MSVS.
>
> Anyone got an answer for this?
See http://www.openoffice.org/issues/show_bug.cgi?id=24588
Fascinating! - and disturbing. I had no idea open source OOo had components that depended on a commercial distribution.
Why disturbing? After all, even to *run* OOo for Windows you need closed-source commercial software (i.e. Windows).
The ability to build and run OOo using all-free software under Linux demonstrates that OOo is not dependent on a commercial distribution.
Vriendelijke groet,
Simon Brouwer --> nl.openoffice.org <--
Open Source multi-platform projects have always striven to at least build with non-commercial components on any platform, commercial or not. It is not part of the technical definition of Open Source, but has always been a primary goal.
I've some - umpf about 15 years - experience with OpenSource and no, this was certainly not a primary goal all the time. It's nice when it's possible but the primary goal was mostly to get the stuff build with the major tool chain on that platform, be it commercial or not. Cygwin and mingw are quite recent additions to the gcc platforms. The StarOffice/OpenOffice code base was around for years before cygwin/mingw became a viable alternative to the Visual Studio (and if you ask the OOo developers the majority of them will assert that at least the Visual Studio debugger is still far superior than gdb)
Yes, I know what you mean. Having worked with various open source projects for almost 30 years I've accepted that there are always people who try to deny or erase the history of open development tools on Windows simply because it's a MS platform. However, as much as I love Linux, since Windows is still the most common end-user OS, the need for open builds is still as important.
This allows users of a commercial OS to build the project without any other product purchase. Many institutions, like schools and NPOs, can only use whatever OS was donated to them or is useful in their vocational training. They must rely on GNU tools for their builds.
Remember it's 'free' as in speech not 'free' as in beer. If these instutions can't afford a commercial compiler they shouldn't use a commercial platform like Windows. Linux is better suited if they are cash strapped. BTW, this is a bogus argument because as far as I can see pratically no one in shools or NGO's will ever build OpenOffice for themselves. It's a daunting experience and only very few people are able and and willing to do the build on Windows, and I bet almost all of them are reading this list :-) Volker and some others have worked very hard to make the build experience a bit more pleasant but it remains a huge pile of unwieldy source code and no amount of work will change that.
Heiner
This is why I alway politely ask first rather than throw terms like "bogus" around. This question was brought to me by a group of high school students. Their school uses donated equipment that came with Windows, and Linux is of no use in their vocational studies since Linux is mostly used on the server side and not in office environments. These students thought that some work on the OOo project would do nicely for a school exhibit at the County Fair science hall on the benefits of Open Source to the general community.
I didn't meant to offend someone. Sorry if I came over a bit to brusk. It's just that porting OpenOffice to mingw is such a major task and while people do see the benefit it's really a question if such a port is really used often enough to make the effort worthwhile. It's not just the one time port, though. This is a 11.5 million lines beast and you have to maintain the port in an ongoing manner to keep it viable. It's not that we didn't investigate such a port. Vladimir Glazounov of StarOffice/OpenOffice release engineering has donated quite a bit of time and others, too. Finally it always comes down to this: If there is someone to do and maintain the port then there will be such a port. If not, well, then not. Well, it's OpenSource, isn't it? Release engineering will, as usual, accept patches if they don't endanger the other platforms.
We (that is, StarOffice/OpenOffice release engineering) did a similiar port once for gcc on Solaris. It has gone stale because of a serious lack of interest.
Just to give you a short overview what it would take to do the port: You've got to compile the 95% of OOo which is identical on all platforms. That's easy because we use gcc on Linux x86, Linux Sparc, Linux PPC, FreeBSD, NetBSD, MacOSX etc in no particular order. The remaining 5% involves creating headers for all the Windows API's used in OOo which still have not a mingw counterpart (I don't know how complete mingw is nowadays, it's some time since I checked). Fix the reliance on Visiual Studio in the Windows specfic part of OOo. Find a way to introduce the quite special Unix/Windows mixed flavour tool chain of mingw into the build system. And finally do some serious assembly level programming for the UNO bridge. That's probably the hard part. It's doable for sure. But it will cost a lot of time and someone has to have a real interest in doing this because otherwise she/he can spend the time better for other things.
It's too bad their simple and polite question has been treated is such a "bogus" manner. I've given them a list of current Open Source multi-platform projects that do not use commercial software their school can not afford, and they've decided that OOo will not be the project to benefit from their efforts. Oh well, I tried.
As much as I hate to say it, OOo is probably not the place to get started with developing (OpenSource or otherwise) software in school. To huge. To diverse. To complex. It changes with a very fast pace. It takes a year or so for an experienced developer to find his/her way around in the source code. It has pieces of wonderful crafted code in it and a few pieces of legacy code you don't really want to show students :-). Ken's suggestion is the right one. It's easier to contribute to OOo by writing components in Java or python.
However, thanks to those who let me know about the current mingw efforts. Even if the kids have chosen other projects, I'll still be tinkering with this one. It's a nice challenge.
Heiner
-- Jens-Heiner Rechtien OpenOffice.org release engineer
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
