[tex-live] stripped binaries? (cygwin 64)

David Carlisle d.p.carlisle at gmail.com
Mon Mar 14 11:52:13 CET 2016


For various reasons I've been building texlive from an rsync'ed copy
of the texlive svn head and also luatex from the luatex svn
experimental branch.

Both seem to work but I notice that luatex.exe is 4 times larger if
built from the texlive tree

-rwxr-xr-x 1 davidc Domain Users  9425920 Mar 13 13:12
./build/texk/web2c/luatex.exe
-rwxr-xr-x 1 davidc Domain Users 41666668 Mar 12 17:04
/usr/local/texlive/2015/bin/x86_64-cygwin/luatex.exe


There are of course some minor differences as the top one is 0.89.5
experimental and the bottom one is .89.4 but the main difference seems
to be that the first one is stripped.

 file reports

./build/texk/web2c/luatex.exe:                        PE32+ executable
(console) x86-64 (stripped to external PDB), for MS Windows

/usr/local/texlive/2015/bin/x86_64-cygwin/luatex.exe: PE32+ executable
(console) x86-64, for MS Windows


In both cases I used the supplied build script to configure/make

so the luatex experimental built with build.sh from:

https://foundry.supelec.fr/scm/viewvc.php/branches/experimental/?root=luatex

and the texlive svn one built with Build from

https://www.tug.org/svn/texlive/trunk/Build/source/



These binaries are only used for internal latex testing so it doesn't
matter too much
if I have a standard build, but so that I can trust the results of
these tests, do these results seem reasonable or have I built one or
the other system incorrectly?

David


More information about the tex-live mailing list