This begs a larger question ... should we just specify iconv as a
requirement for the next release?
You're thinking those without it, e.g. Andy, could easily get iconv with
a package install?
Yeah ... I am kind of surprised OpenBSD doesn't ship with iconv, but I
have to believe it's available via a package. Like I said, it's part
of POSIX so I think requiring it is perfectly reasonable. Maybe he does
have it and our autoconf tests are wrong; if that's the case we should
spend our energy on fixing that.
I'm not sure it makes sense to have all of those #ifdef HAVE_ICONV
directives cluttering up the code,
Well, there's not too many, and a few of them could be simpler to
understand if they didn't try to thread through a function but instead
just got the no-iconv case over with by returning, leaving the rest in
the #else, even if it means a little duplication, e.g. the return
statement.
Well, I think we have different definitions of "not too many", but
regardless ...
- As we've seen many times, it's easy to get the character conversion code
wrong, and it's kind of important we get it right. Having those #ifdefs
makes it harder to read. I mean, yes, we could restructure that all,
but I have to ask ... why?
- Given the reality of email in 2018 means there's a high probability you're
going to be getting stuff with a character set different than your native
one, I have to wonder how useful nmh is without iconv support.
- I know it's possible to test everything WITHOUT iconv support, but it
seems to me that to do that properly would require us to build nmh without
iconv support and run it through the test suite, but that just seems
like a waste of energy.
- My eventual goal is to rototill the MIME support and I am not interested
in the least in maintaining support for systems without iconv.
In theory we should be testing things without iconv support to catch
these problems, but I don't think that's worth the time.
I'm doing that. I ./configure'd like normal and then s/1/0/ in Makefile
and config.h for HAVE_ICONV. All went well with `make all', one test
was skipped with `make check', but one test failed in
test/mhshow/test-charset, and I'm not sure why.
[...]
I think this just bolsters my arguments :-)
--Ken
--
Nmh-workers
https://lists.nongnu.org/mailman/listinfo/nmh-workers