opensubscriber
   Find in this group all groups
 
Unknown more information…

g : gs-devel@ghostscript.com 23 December 2009 • 2:14AM -0500

[gs-devel] assert vs debuginfo
by Ralph Giles

REPLY TO AUTHOR
 
REPLY TO GROUP




Tim,

We're considering (re-)adding assert calls to the ghostscript library.
The general idea is to default to having them on in debug builds, and
off in release builds. (If you're fuzzy on the rules, as I was,
assert() calls abort() when a condition fails unless the magic NDEBUG
cpp symbol was defined when assert.h was included...and library's
calling abort() results in a poor user experience).

I'm curious how this will affect packagers, and in particular the
debuginfo packages fedora provides. I couldn't find much documentation
on that, but I gather the recommended practice to to hack -g into the
release build flags, and then rpmbuild produces both a stripped
executable and a debuginfo package. Does this also work on libraries?
Will however you're hacking in the -g flag maintain the NDEBUG define?

-r
_______________________________________________
gs-devel mailing list
gs-devel@ghos...
http://ghostscript.com/cgi-bin/mailman/listinfo/gs-devel

Bookmark with:

Delicious   Digg   reddit   Facebook   StumbleUpon

Related Messages

opensubscriber is not affiliated with the authors of this message nor responsible for its content.