I'm not very clear on the architecture of the current texi2any in its
various configurations, but I want to point out a significant
advantage of a C-only implementation, as opposed to Perl with C
extensions -- the fact that the former produces binary distributions
that are not bound to a specific version of Perl.  If one needs to
install Texinfo on a system other than where it was built, or
distribute precompiled binaries (as, for example, I do on the
ezwinports site), the fact that the extensions need a particular
version of Perl is a huge disadvantage.  It means that one either
needs to distribute the whole Perl source tree (to be consistent with
GPL), or one needs to remove the XS extensions from the distro, thus
making texi2any much slower.

AFAICS, Texinfo 7.2.90 adds quite a few new shared libraries, but they
all depends on the Perl library.

So if ctexi2any allows, or could allow in some not-too-distant future,
to run the Texinfo converters without any need for Perl, that would be
a significant advantage, even if speed-wise it is not much faster.

Thus, I think that developing Texinfo in the direction that will allow
eventually to use Perl only for optional extensions is a very good and
worthy goal.

Of course, I'm not going to do much of this work, so my opinions
should be taken with a grain of salt.

Thanks for listening.

Reply via email to