During the build following sed script is being used:
[snip]
cp /dev/null sedscript
for f in ../kpathsea/paths.h; do \
sed -n -e '/^#define/s/#define[ ][
]*\([A-Z_a-z][A-Z_a-z]*\)[ ][ ]*\(.*\)/s%@\1@%\2%/p' \
$f \
| sed -e 's/"//g' -e 's/[ ]*\/\*[^*]*\*\///g' >>sedscript;\
done
[/snip]
Unfortunately [A-Z_a-z] under et_EE locale this range is not the same range as
it is wanted (Estonian alphabet: ..p,q,r,s,š,z,ž,t,u,v,...) and this causes
build error:
[snip]
tex-file.c:438: error: 'DEFAULT_GFFONTS' undeclared (first use in this
function)
tex-file.c:438: error: (Each undeclared identifier is reported only once
tex-file.c:438: error: for each function it appears in.)
tex-file.c:445: error: 'DEFAULT_PKFONTS' undeclared (first use in this
function)
tex-file.c:452: error: 'DEFAULT_GLYPHFONTS' undeclared (first use in this
function
...
[/snip]
Typical shell-scripting bug... looks like this is an upstream problem, but
gentoo tex maintainers should be able to patch it for now and help upstream to
fix it. Could you help them find the problem more quickly by specifying which
file has the broken code you posted?
thanks for the report, i've set LC_ALL to C at the top of the src_compile.
After grepping the source, there are waaaaayyy too many A-Z a-z regexps... even
if we fix the build problem, there may very well remain some hidden problems,
so i've prefered the simple way of setting C locale instead.