A list for the developers of CellML tools

Text archives Help


[cellml-dev] Compiling the API and OpenCell on Windows


Chronological Thread 
  • From: r.britten at auckland.ac.nz (Randall Britten)
  • Subject: [cellml-dev] Compiling the API and OpenCell on Windows
  • Date: Thu, 18 Nov 2010 22:00:43 +1300

Yes, not sure anyone has taken that approach before, but in theory, it
should work.

> -----Original Message-----
> From: cellml-tools-developers-bounces at cellml.org [mailto:cellml-tools-
> developers-bounces at cellml.org] On Behalf Of Lucian Smith
> Sent: Thursday, 18 November 2010 8:56 p.m.
> To: A list for the developers of CellML tools
> Subject: Re: [cellml-dev] Compiling the API and OpenCell on Windows
>
> Oh, excellent! I presume this is a 32-bit msvc build?
>
> -Lucian
>
> * Randall Britten <r.britten at auckland.ac.nz> [2010-11-18 06:42] writes:
> > Yes, we will be providing binaries, and yes, grabbing the binaries from
> > OpenCell's installer should work. The OpenCell installer can actually
> just
> > be unzipped (7-zip does this nicely).
> >
> > Regards,
> > Randall
> >
> > > -----Original Message-----
> > > From: cellml-tools-developers-bounces at cellml.org [mailto:cellml-
> tools-
> > > developers-bounces at cellml.org] On Behalf Of Lucian Smith
> > > Sent: Thursday, 18 November 2010 7:21 p.m.
> > > To: A list for the developers of CellML tools
> > > Subject: Re: [cellml-dev] Compiling the API and OpenCell on Windows
> > >
> > > Actually, don't I remember reading in the minutes or something that
> you
> > > were considering having a binary distribution of the code? I
> downloaded
> > > OpenCell itself; if I used that, along with a snapshot of the code
> from
> > > when it was created for #include purposes, should everything work?
> > >
> > > -Lucian
> > >
> > > * Randall Britten <r.britten at auckland.ac.nz> [2010-11-18 01:33]
> writes:
> > > > Hi Lucian
> > > >
> > > > Andrew is quite busy with his PhD at the moment, so I think you
> will
> > > only
> > > > get a comprehensive reply from Andrew in a day or perhaps a few
> days,
> > > but I
> > > > had a brief mail discussion with him, and he suggested a few
> > > alternatives
> > > > that you could try:
> > > >
> > > > 1) Use the setup tool he created for setting up a windows
> environment
> > > for
> > > > building the CellML-API:
> > > > http://cellml-api.hg.sourceforge.net/hgweb/cellml-api/win32-build-
> > > setup/.
> > > > Note that it is still a bit rough, and requires more testing, so if
> you
> > > do
> > > > try this, be prepared for retrying a few times, probably preferable
> if
> > > you
> > > > have pristine VM images on stand-by. If you do, it will help us
> finish
> > > it
> > > > off, which will make life easier for developers in future.
> > > >
> > > > 2) If you want to avoid using MSVC, and only use cygwin, then you
> will
> > > have
> > > > to build xul-runner on cygwin from source, since Windows xul-runner
> > > binaries
> > > > distributed by Mozilla are usually built using MSVC.
> > > >
> > > > 3) Change your app to use the new "TeLICeMS" (available in "head
> > > revision"
> > > > of source code repo), which will remove the xul-runner dependency.
> The
> > > work
> > > > on xul-runner support was part of the OpenCell project, but our
> group
> > > is no
> > > > longer working on the OpenCell project, and the new OpenCell
> project is
> > > not
> > > > based on xul-runner. So this is probably the best long term option,
> > > since
> > > > it means improves maintainability of the CellML support in your
> > > application
> > > > over the next few years. A pain in the short term though.
> > > >
> > > > TI's relevant to TeLICeMS:
> > > > https://tracker.physiomeproject.org/show_bug.cgi?id=1356
> > > > and
> > > > https://tracker.physiomeproject.org/show_bug.cgi?id=2777
> > > >
> > > >
> > > > Regards,
> > > > Randall
> > > >
> > > >
> > > >
> > > > > -----Original Message-----
> > > > > From: cellml-tools-developers-bounces at cellml.org [mailto:cellml-
> > > tools-
> > > > > developers-bounces at cellml.org] On Behalf Of Lucian Smith
> > > > > Sent: Thursday, 18 November 2010 1:33 p.m.
> > > > > To: A list for the developers of CellML tools
> > > > > Subject: [cellml-dev] Compiling the API and OpenCell on Windows
> > > > >
> > > > > So I really really tried to follow the Windows build instructions,
> > > > > despite
> > > > > the fact that they are ridiculously arcane, and that they are
> > > designed to
> > > > > install in c:/, of all places. I made a cellml-specific version
> of
> > > > > Cygwin
> > > > > (since the instructions morph the install into something
> completely
> > > > > unuseable by anybody else), and at first I tried to see if I
> could
> > > get it
> > > > > to work with the most recent versions of automake, etc, since
> really,
> > > > > what
> > > > > the heck. It took forever, and in the end I was not able to get
> > > > > build/msvc-wrapper to run at all.
> > > > >
> > > > > So, I thought to myself, "Well, maybe it really does need the
> ancient
> > > > > automakes, etc." so I went through and crippled it, and in the
> end, I
> > > > > typed 'aclocal' and it said, "wtf I can't use these ancient
> versions
> > > of
> > > > > automake!" so I gave up. I figured it was only a matter of time
> > > until
> > > > > some other dependency incompatibility was discovered.
> > > > >
> > > > > So I took a different tack: I tried compiling an honest-to-
> goodness
> > > > > Cygwin compile, instead of the weird cygwin/visual studio hybrid
> > > > > described
> > > > > in the instructions. My own software compiles on Cygwin, so I
> > > figured
> > > > > since the CellML API was really developed for unix anyway, I
> could
> > > > > probably get that to work more straightforwardly.
> > > > >
> > > > > It configured OK (with some warnings I'll talk about in a second)
> but
> > > ran
> > > > > into troubles with xulrunner. I get the error:
> > > > >
> > > > > $ make
> > > > > mkdir -p ./simple_interface_generators/glue/xpcom
> > > > > /home/Lucian/CellML/hg/cellml-api/./build/cygwin-wrapper
> > > > > /home/Lucian/xulrunner-sdk/bin/xpidl -m header -
> > > I/home/Lucian/xulrunner-
> > > > > sdk/idl -e simple_interface_generators/glue/xpcom/IWrappedPCM.h
> > > > > simple_interface_generators/glue/xpcom/IWrappedPCM.idl
> > > > > .\simple_interface_generators/glue/xpcom/IWrappedPCM.idl:2: can't
> > > open
> > > > > included file nsISupports.idl for reading
> > > > >
> > > > > This despite the fact that /home/Lucian/xulrunner-
> > > sdk/idl/nsISupports.idl
> > > > > exists. So I'm not sure what's going on here.
> > > > >
> > > > > Then I tried configuring it without xulrunner, and had much
> better
> > > luck.
> > > > > There were the ominous warnings from 'configure':
> > > > >
> > > > > checking libxml/tree.h usability... yes
> > > > > checking libxml/tree.h presence... no
> > > > > configure: WARNING: libxml/tree.h: accepted by the compiler,
> rejected
> > > by
> > > > > the preprocessor!
> > > > > configure: WARNING: libxml/tree.h: proceeding with the compiler's
> > > result
> > > > > checking for libxml/tree.h... yes
> > > > >
> > > > > and
> > > > >
> > > > > checking dlfcn.h usability... yes
> > > > > checking dlfcn.h presence... no
> > > > > configure: WARNING: dlfcn.h: accepted by the compiler, rejected
> by
> > > the
> > > > > preprocessor!
> > > > > configure: WARNING: dlfcn.h: proceeding with the compiler's
> result
> > > > > checking for dlfcn.h... yes
> > > > >
> > > > > And indeed, it compiled a few things until it tried to include
> > > > > libxml/tree.h and couldn't find it. I was able to fix this by
> > > editing
> > > > > the
> > > > > makefile from
> > > > >
> > > > > LIBXML_CFLAGS =
> > > > >
> > > > > to:
> > > > >
> > > > > LIBXML_CFLAGS = -I/usr/include/libxml2
> > > > >
> > > > > The problem here was that the '-I' bit was mistakenly placed in:
> > > > >
> > > > > CFLAGS = -I/usr/include/libxml2 -O2 -ffast-math
> > > > >
> > > > > which wasn't being used in that place.
> > > > >
> > > > > Then it complained about a nonexistent library 'stlport.5.0'.
> This I
> > > > > fixed by changing
> > > > >
> > > > > STLLINK = -lstlport.5.0 -no-undefined
> > > > >
> > > > > to
> > > > >
> > > > > STLLINK = -lstdc++ -no-undefined
> > > > >
> > > > > At that point, the whole thing actually compiled! But still
> without
> > > > > xulrunner. And with the slightly ominous warnings all over the
> place:
> > > > >
> > > > > libtool: link: warning: `/usr/lib/gcc/i686-pc-
> > > > > cygwin/4.3.4/../../../libxml2.la' seems to be moved
> > > > >
> > > > > despite the fact that it was right there (/us/lib/libxml2.la)
> > > > >
> > > > > However! I did a 'make install' into a test directory, and after
> I
> > > > > copied
> > > > > in some cygwin dlls, the test exe's actually ran! I about fell
> over.
> > > > >
> > > > > So at this point, all I need is a way to get the xulrunner-sdk to
> > > work
> > > > > under cygwin for your build process. Any ideas? Thanks!
> > > > >
> > > > > -Lucian
> > > > > _______________________________________________
> > > > > cellml-tools-developers mailing list
> > > > > cellml-tools-developers at cellml.org
> > > > > http://lists.cellml.org/mailman/listinfo/cellml-tools-developers
> > > >
> > > > _______________________________________________
> > > > cellml-tools-developers mailing list
> > > > cellml-tools-developers at cellml.org
> > > > http://lists.cellml.org/mailman/listinfo/cellml-tools-developers
> > > _______________________________________________
> > > cellml-tools-developers mailing list
> > > cellml-tools-developers at cellml.org
> > > http://lists.cellml.org/mailman/listinfo/cellml-tools-developers
> >
> > _______________________________________________
> > cellml-tools-developers mailing list
> > cellml-tools-developers at cellml.org
> > http://lists.cellml.org/mailman/listinfo/cellml-tools-developers
> _______________________________________________
> cellml-tools-developers mailing list
> cellml-tools-developers at cellml.org
> http://lists.cellml.org/mailman/listinfo/cellml-tools-developers





Archive powered by MHonArc 2.6.18.

Top of page