John Levon wrote:
> On Wed, Nov 09, 2005 at 11:16:32AM +0000, Angus Leeming wrote:
>
>> Sorry, that's FUD. AC_MSG_ERROR (currently disabled) would print an
>> error message and abort. LYX_ERROR would not abort but would print an
>> error message at the end of configuration. In other words, it would
>> do exactly what the current code in qt.m4 does together with the
>> cobbled on bit at the end in configure.ac. The difference is that,
>> with my suggestion, the code is self-contained.
>
> I suppose I should mention again that I have no idea why it's a good
> idea to have to wait to the end to discover your machine setup is
> screwed.
My turn to be baffled. If you're watching configure do its stuff you'd see
something like:
Looking for Qt library... failed
With your original, FATAL=0, code you'd get no other error message at all.
At least with Georg's addition to configure.ac, we get something
reasonably descriptive at the end of the configure process as well. I'm
simply saying that we should go the whole hog, use FATAL=1 but, rather
than terminate the configure run at AC_MSG_ERROR, use LYX_ERROR to give us
a clear error message at the end together with encapsulated code. Why the
opposition?
--
Angus
