lilypond-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Fwd: GDP Docs compilation FAILED (2011.09.23-03:32)


From: David Kastrup
Subject: Re: Fwd: GDP Docs compilation FAILED (2011.09.23-03:32)
Date: Fri, 23 Sep 2011 09:55:26 +0200
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.0.50 (gnu/linux)

David Kastrup <address@hidden> writes:

> Peekay Ex <address@hidden> writes:
>
>> Hello again..
>>
>> On Fri, Sep 23, 2011 at 5:34 AM, Peekay Ex <address@hidden> wrote:
>>> Hello,
>>>
>>> On Fri, Sep 23, 2011 at 3:46 AM, Reinhold Kainhofer
>>> <address@hidden> wrote:
>>>> Am Friday, 23. September 2011, 04:35:06 schrieben Sie:
>>>>> > fatal error: failed files: "be/lily-48175def.ly"
>>>>>
>>>>> A decent build system would take that filename, grep it for
>>>>> "\sourcefilename", and tell us *that* information, instead of the
>>>>> current useless output.  :)
>>
>> I used gitk to roll back the head to David's first branch merge
>>
>> http://git.savannah.gnu.org/gitweb/?p=lilypond.git;a=commit;h=31dc6e0a1b454c1186052f36cfac3e2e7c20325e
>>
>> and a full make doc (from a fresh build dir) worked fine.
>>
>> So the issue, by deduction, is somewhere in his second merged branch
>> (Merge branch 'musicfunction-optional-arguments' -
>> 83055a30e52c14b0fd49d6df3eb1c7af476ecb4b) and/or onwards.
>>
>> Hope that helps.
>>
>> Sorry David :(
>
> Would have been too lucky.  Sigh.  Looks like the rebasing managed to
> make a non-working version out of several working (and regtest passing)
> patches.
>
> I'll try fixing this in the next hour, otherwise I am going to revert.
> Documentation is in the queue close to be checked in, specific regtests
> will follow.

Here is the problem: \with-link declares its argument type wrong
(declares symbol? and uses #"label", a string).  My patch has the
side-effect that markup commands actually check their argument lists for
correctness.  Previously it just let any Scheme argument through
completely unchecked.

Now I could revert my branch merge, or I could prepare a patch that
stops type-checking the argument lists to markup commands, but I think
it is saner if we grit our teeth and just fix the previously unnoticed
bugs as they get exposed.

Sorry about that.  I am fixing the one with \with-link in the next 15
minutes.  Given my slow computer, it might help if others do the regtest
runs after that.

I am somewhat surprised that the patch passed testing by someone else,
though.  It would seem that doc builds are not part of that.

-- 
David Kastrup




reply via email to

[Prev in Thread] Current Thread [Next in Thread]