Buildfarm member hamerkop has a niggle about this patch:
c:\\build-farm-local\\buildroot\\head\\pgsql.build\\contrib\\fuzzystrmatch\\daitch_mokotoff.c
: warning C4819: The file contains a character that cannot be represented in
the current code page (932). Save the file in Unicode format to prevent data
loss
It's complaining about the comment in
static const char iso8859_1_to_ascii_upper[] =
/*
"`abcdefghijklmnopqrstuvwxyz{|}~ ¡¢£¤¥¦§¨©ª«¬
®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ"
*/
"`ABCDEFGHIJKLMNOPQRSTUVWXYZ{|}~ !
?AAAAAAECEEEEIIIIDNOOOOO*OUUUUYDSAAAAAAECEEEEIIIIDNOOOOO/OUUUUYDY";
There are some other comments with non-ASCII characters elsewhere in the
file, but I think it's mainly just the weird symbols here that might fail
to translate to encodings that are not based on ISO 8859-1.
I think we need to get rid of this warning: it's far from obvious that
it's a non-issue, and because the compiler is not at all specific about
where the issue is, people could waste a lot of time figuring that out.
In fact, it might *not* be a non-issue, if it prevents the source tree
as a whole from being processed by some tool or other.
So I propose to replace those symbols with "... random symbols ..." or
the like and see if the warning goes away. If not, we might have to
resort to something more drastic like removing this comment altogether.
We do have non-ASCII text in comments and test cases elsewhere in the
tree, and have not had a lot of trouble with that, so I'm hoping the
letters can stay because they are useful to compare to the constant.
regards, tom lane
