Re: [lmi] PATCH: Switch to using std::filesystem

[Vadim Zeitlin]

On Thu, 29 Apr 2021 16:38:10 +0000 Greg Chicares <gchicares@sbcglobal.net> 
wrote:

GC> Now that I've studied the first few commits, the real surprise is
GC> that you and Ilya finished before the end of this decade. The
GC> filesystem part is almost simple; coping with the still-evolving
GC> utf-8-ization of C++ is a Herculean labor.

 I don't think it's that bad in general, but I was really surprised that
C++20 changed std::path in backwards incompatible way. I'm not used to
seeing changes like this in C++ evolution.

 Concerning UTF-8-ization, I'll make another patch soon adding plenty of
missing conversions to/from UTF-8. I know that you don't care much about
supporting non-ASCII characters (unless it has changed since the last time
we've discussed it), but I think we really need to handle them correctly
even when not dealing with foreign languages because they very commonly
occur even in English texts nowadays (e.g. they do appear regularly in
this mailing list archives) and we really don't want to lose the entire
string contents just because it contains a "smart quote" or something like
this in the middle (but this is what happens now).

 In fact, my goal is to define wxNO_IMPLICIT_WXSTRING_ENCODING for lmi
build to ensure that an explicit encoding is always specified when
converting between Unicode and narrow strings (and, of course, this
encoding will always be UTF-8 in lmi). Actually enforcing this will have to
wait for the next wx update, but I can already make the required changes to
prepare for it.

GC> Okay...I reenable my brutal workaround from yesterday (i.e.,
GC> blocking only one pair of statements in one unit test), and
GC> every known test passes, so I'll push it now and spend a few
GC> more hours studying the details (and updating our proprietary
GC> product repository) later.

 Thanks a lot for merging it so quickly!
VZ

pgpcqetrV9VIW.pgp
Description: PGP signature