Re: texi2dvi: More functions

[Akim Demaille]

Le 20 janv. 05, à 20:25, Karl Berry a écrit :

> Hi Akim,
Hi Karl,

>     reading $TEXINPUTS helped
>
> I have no problem with reporting $TEXINPUTS after absolutification, but
> I have some questions about other parts of the patch:
>
> (1) do we really gain that much by making verbose a function?  I guess
>     $0 might be helpful in the output, although you could also argue
>     that it is additional noise.

Actually, I should have reduced it to basename.  But since it is
uncommon to run texi2dvi by hand, and common to have it part of
a Makefile, or script, I think it's better to sign the messages.
But I won't make a big fuss about this :)

> (2) I'm not sure verbose output should go to stderr.

Hm, I do.  Tracing is not actual output.  Anything that is not
actual output should go onto stderr IMHO.  Including for programs
that have no actual stdout output, such as texi2dvi.

> (3) I was under the possibly erroneous impression that
> funcname ()
> {
>   is not sufficiently portable.
> }

On Libtool they made the same move, so I guess it is sufficiently
portable.  What is not is using the keyword function.

>   that is, the { has to be on the same line (or the newline escaped by 
> a \).
>   I'm sure this was true at one time, although I can't say any more 
> what
>   version of what system I was using.

ISTR that texi2dvi, before my change, already had functions
declared this way.  I don't really care, if you prefer
the other one, feel free to make it uniform the other way.