Re: Compile issues

[Daniel Pfeiffer]

Saluton Dave,

Dave Love <address@hidden> skribis:

> address@hidden (Daniel Pfeiffer) writes:
> 
> > We are currently thinking about various remedies for this.  In the
> > mean time start with M-x first-error.
> 
> I'm surprised it was installed if the normal user interface doesn't
> work.

"Doesn't work" is an exaggeration.  Personally I tend to scroll around and
click with the mouse.  But I had extensively tested the keyboard commands, the
way I thought they'd be used.  For this I mostly used the new feature of being
able to save a compilation buffer.  When visiting such a file, I would of
course always be on the first line, so I never noticed.  And my changes had
been published for several months without any of the several users noticing
this.

> >> NEWS reports that `by default' warning messages are skipped, but gives
> >> no indication of how to get the old behaviour back.  I think the old
> >> behaviour should be the default.
> >
> > No way!  It states that only informationals are skipped.
> 
> OK, but I assumed it was lumping `information' and `warning' together,
> since it's not clear to me that there's a useful distinction in
> practice.  For instance, what would be warnings from GCC may appear as
> `Info' from the DEC^WCopmpaq^WHP compilers though they're things you
> need to see when trying to fix Emacs on Tru64.

A message labeled "info" is a clear statement that this is not something to
worry about.  Of course with rotten languages like C++, which gets stricter
every year, you're asking for trouble if you use an old compiler.

Others (e.g. at my work) lump warning and error together, saying a program may
have neither.  Otoh Sun's javac makes catching an exception which can't occur
there an error — Imho that's at best worth a warning.  But some compilers
makes these distinctions and I find it useful to reflect them!

> Anyhow, if the facility to match them is there, it's not clear how to
> enable it.

It's in NEWS now.  Btw. Emacs can customize groups, and then you can
comfortably browse all the options.

> > It is only slightly incompatible.  Mostly tacking an '^' to the beginning
> > of your regexps should be enough.
> 
> I think that should be documented.

It was of course documented, now it's also mentioned as a change in NEWS.

> What feature should external
> packages test to know what version of messages they should use when
> they have to support the released and new version?

There is no specific feature, but you can test the existence of any of the new
variables or functions (see Changelog).

> > If you call a function called *-internal, you're bargaining for trouble
> > :-)
> 
> Yes, but.  Several different packages in Emacs did that.  It appeared
> to be sanctioned for that reason.  If it was wrong, one would expect
> the first Emacs package that needed to do it to fix the issue.

Some of them I can't test or do not even understand.  Their authors and
maintainers have repeatedly been asked to cooperate, but did not respond.  So
we decided to install this, so that someone who finds something broken, can at
least explain that.

coralament / best Grötens / liebe Grüße / best regards / elkorajn salutojn
Daniel Pfeiffer

-- 
lerne / learn / apprends / läramå / ucz się    Esperanto:
                              http://lernu.net/