Re: please do a "make doc-clean; make clean"

[Carl Sorensen]

On 12/30/09 3:14 PM, "Graham Percival" <address@hidden> wrote:

> On Wed, Dec 30, 2009 at 08:31:11AM -0700, Carl Sorensen wrote:
>> 
>> I did end up doing a reconfigure (due to trying the out-of-tree build, which
>> also failed, but that's another story).
>> 
>> During my configure, I've found the following warnings:
>> 
>> checking g++ version... 4.0.1
>> configure: WARNING: autoconf <= 2.59 with g++ >= 3.3 gettext test broken.
>> configure: WARNING: Trying gcc, cross fingers.
> 
> That's normal, and has been for years.  (IIRC, we even have an
> open issue for "clean up the configure so people don't get
> confused by pointless warnings")

That was my memory.

> 
>> checking Python.h usability... yes
>> checking Python.h presence... no
>> configure: WARNING: Python.h: accepted by the compiler, rejected by the
>> preprocessor!
> 
> That's not normal.

How can I find out about the tests used to check "usability" and "presence"?
I would like to know what exactly is failing.  That might give me a clue.

Every one of the strange warnings says that *.h is usable, but not present.
That seems strange.


> 
> Well, the problem is either in the git source, or something
> special on your system.  The last change to configure.in was
> removing some old pango stuff in Nov, which I can't imagine
> causing this.
> 
> So what happened recently on your system?  Did you move the
> lilypond git dir onto a removable HD?  I had some weird problems
> with configure, but it turned out that I'd put it on an SD card
> and the whole drive was marked noexec.  (security issue)

I downloaded some Mac system updates (and make worked fine after
I did so).

I haven't done anything with the git sources.

I did decide to try the out-of-tree build, which required me to do a new
configure.  I thought I could do a configure in the out-of-tree build
directrory, but the error messages said I had to do a make distclean first.
After doing make distclean, configure doesn't seem to work any more.


> 
> Did you update macports?  A new version of gcc, or make, or
> automake, or autoconf, or whatever, could cause it.

When I did the ivo boumans build instructions to try to build 2.12 using
macports, it updated a *bunch* of stuff, but it made no difference.  Neither
2.12 nor 2.13 will build now.
> 
> Did you change your .profile ?  Maybe you added
>   export PATH=/opt/new/version/of/matlab
> instead of
>   export PATH=/opt/new/version/of/matlab:$PATH

No, no changes to profile.  My paths seem to be OK.


> 
> If none of those turn up anything... well, is your HD dying?
> Apparently the SMART disk checker (in OSX, it's in the disk
> utility) isn't terribly reliable; google found that a third of
> their HDs died without any warnings (and nobody on earth who is
> allowed to publish unclassified research papers uses more HDs than
> google).  But hey, that still gives a 2/3 chance that if there
> _is_ someting wrong, it'll show up.

Disk checker shows no problems.



> 
> 
> After all, once we've eliminated the impossible, the only thing
> left, however improbable, is the cause.  Something like that, at
> least.
> 
> Cheers,
> - Graham