debug and release? about afdko HOT 5 CLOSED

Why? Habit, and a lack of knowledge about Linux development.When I run the build scripts, I usually want a release version from which I can make a working distribution. If you can add this to the scripts, with an option to make the debug versions instead, feel free to do so.

from afdko.

Putting a -g then strip, is not the same as doing without -g - bugs in compiler and and strip aside. There are bugs in compilers from time to time where only the result with -g or without it works, and there are the occasional bugs with strip also, where applying strip actually cripples your binary. Those are side issues.

Anyway, the difference, in principle, is that if you do -g, you think there is a possibility the outcome might be broken and you intend to look into it. -g disables some optimizations which might make debugging impossible, as a general rule. Major inlin'ing, code re-arrangements, etc. So as a general rule, the resulting binary from -g then strip will still be slower than without -g. (assuming your compiler is working correctly in both cases).

That's why, depends on circumstances, you want to do both, or either.

from afdko.

Alternatively, in windows' land, it is routine to ship only "release" binaries, for exact this reason: to make running and poking into the binary with a debugger difficult or impossible. On the typical linux box, debug info are stripped, but stored externally, so obviously it is still possible.

from afdko.

In fact -g then strip is perhaps worse than either plain -g or without -g: you don't get a fully optimized binary, and you cannot debug with it either.

I'd say, if you have to choose one, make up your mind which is more important - fully speed or debugg'ability.

from afdko.

Fixed via #414

from afdko.