Le 09/09/2020 à 11:00, Enrico Forestieri a écrit :
Do you have a concrete case in mind where deriving requirements from the
contents of the macro goes wrong? Note that in this case, we could decide
that the requirements given to the macro override whatever is decided from
its contents.
The point is that the macros defined in the symbols file are only used
for on-screen representation. Their definition will never appear in the
output. So, if \int is defined in the symbols file as
\def\int{\intop\nolimits} esint|amsmath
meaning that it requires either esint or masmath, the requirements for
\intop should not be taken into account because \intop will never
appear in the latex file.
But \intop can appear in a LaTeX file, and it should work. So we have to
set the requirements of \intop anyway.
Here, the definition of \int is only used
for representing it on the screen. For example, if you define
\def\int{\vcentcolon} esint|amsmath
because you like to see \vcentcolon instead of \int on screen, then
you would get the package mathtools as a requirement. But \vcentcolon
does not really appear in the output and that requirement is bogus.
I understand that but:
1/ we do not want to do such bogus things in symbols file, unless you
have a better use case in mind.
2/ this is why I proposed that in the case where requirements are given
for the macro, the requirements induced by the definition are
skipped.
The proposal 2/ is of course not 100% precise, since it is difficult to
make a difference between "no requirement needed" and "no requirement
given".
But in practice it seems easier in terms of maintenance to let LyX infer
the requirements.
JMarc
--
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel
