Re: Help with GUB, or manual cross-compile

[David Kastrup]

Chris Yate <address@hidden> writes:

> Following up a conversation on the user list, as this might be a better
> place to ask... I think we've established that the only way to compile for
> Windows is to use GUB on Linux to cross-compile for mingw. Well, GUB barfs
> all the time here. I could start posting chunks of GUB log messages, but I
> don't feel that's really useful at this point -- if someone is able to help
> diagnose the problem we can look into that!
>
> First, to make sure I'm not wasting my time, can somebody suggest what
> build platform and configuration I need to successfully cross-compile
> Lilypond for Windows? I've tried a number of different Ubuntu versions,
> either VMs or linux running directly on the metal. I've had to install a
> few libraries (mpf, mpc, zlib etc.) from apt in order to get a step further
> in GUB. I've also installed the mingw g++ packages, though I'm not sure
> whether that is relevant.
>
> Also, do I need to anything more than:
>     $ bin/gub mingw::lilypond
> ?
>
> Alternatively, I'm sure it should be possible (possibly easy) to
> ./configure the lilypond source code directly for cross compilation. If
> someone has a quick solution for how to do that, I'd be really grateful of
> it.
>
> Context: I'm getting seriously fed up with a bug in Lilypond that's causing
> the build of scores of larger than smallish size to crash. It's a very
> repeatable assertion failure, "Line 1180 in Page-Breaking.cc". However,
> I've not found it easy to create a reliable "minimal" example for this. It
> only seems to happen in reasonably complicated scores with 10+ staves and a
> few pages.  In any case, unless I can compile this for Windows I wouldn't
> be able to investigate the problem at all.

If you can run this in a debugger in Windows, you could try giving a
backtrace.  It might help enough for getting closer to the source of the
problem.

-- 
David Kastrup