On Wed, Sep 23, 2026 at 08:24:39PM -0400, D wrote: > Hopefully there should be no significant differences between mine and > Jakob's patchset. > > Assuming that Jakob's will be upstreamed over mine, I at least should > be given co-credit, since I worked on this before Jakob submitted his > on the mailing list...
My patch has been committed already, see https://gcc.gnu.org/r17-4409 and the commit log can't be amended anymore. >From what I can see, there are some differences between the patches, so where it is intentional please mail separate incremental patches instead. >From what I can see, you've non-trivially changed libstdc++-v3/include/bits/unicode.h, that change should be posted separately to Jonathan and both [email protected] and [email protected] mailing list with rationale, explaining what if anything has changed in the Unicode algorithm and likely needs to be covered by a testcase. Patches should have a ChangeLog entry in the commit message, see https://gcc.gnu.org/codingconventions.html The libcpp/makeuname2c.cc changes in your patch are incorrect, the idx needs to be unique for every unique prefix, all ranges with say "CJK UNIFIED IDEOGRAPH-" have the same idx but "JURCHEN CHARACTER-" and "NUSHU CHARACTER-" need each a different one. For each new range some test coverage is needed (included in my commit) and that wouldn't pass with your version. I see you've regenerated libcpp/printable-chars.inc and libcpp/combining-chars.inc, I didn't. The reason is simple, contrib/unicode/README doesn't document those. So, the README should be patched first to say to also run those. Or maybe better not, because unlike the other scripts it actually uses python3-libs unicodedata instead of parsing the updated contrib/unicode/*.txt files. That means it actually doesn't update the files to the new Unicode standard, but usually a few years old one (both my python3-libs and what you've posted seems to be Unicode 16, so 2 years old now). I think it would be best to change those scripts to actually parse the files, e.g. for combinindg-chars.inc I think all it cares is about parsing the first (codepoint) and 4th ; separated column of UnicodeData.txt, if the 4th is 0, it is not combining, if it is something else, it is combining. I fail to see the point of the is_combining array, it is always 0, 1 repeated many times, why do we need it? printable-chars.inc similarly, why do we need is_printable and it is 1st and 3rd column, if 3rd column starts with C, it is non-printable, if it doesn't, it is printable. contrib/unicode/gen-box-drawing-chars.py likely doesn't need to be ever rerun because it only looks up the codepoints for a few chars with names and the names and codepoints should be immutable. Jakub
