On Fri, Oct 11, 2013 at 01:38:20PM +0100, Derek Buitenhuis wrote: > On 10/11/2013 12:48 PM, Diego Biurrun wrote: > >>> Then if strip is added back, it should be a proper silent command. > >> > >> What OS specific options? This does *not* strip the libs the way you think, > >> and thus we do not need the OS-specific options, AFAIK. > > > > The command generates warnings on some platforms, presumably because > > the syntax is nonportable. What are those warnings? Does the stripping > > work at all on the platforms that generate warnings? > > +------------------------------------------------------------+ > | Warning: You have not read all emails associated with | > | this commit yet.yet. | > | | > | Somebody else may already have answred what you are about | > | ask in your reply. Please read the all context before | > | replying to any email about it. | > | | > | [CANCEL] | > +------------------------------------------------------------+ > > http://lists.libav.org/pipermail/libav-devel/2013-October/051802.html
That mail only refers to "warnings" and "not working". I have yet to see the actual warnings pasted or get confirmation that the stripping either works as expected or has no effect on those platforms. Now turn the drama control knob down a notch please. > strip is not POSIX. strip is *only* binutils. http://pubs.opengroup.org/onlinepubs/9699919799/utilities/strip.html > The only "portability" check we need is for a sufficiently new > version. Technically, we don't even need that, since as it is, if -wN > is not recognized, it is a noop, which is the correct behavior for our > use case. We are *only* stirpping the spurious labels, and nothing > else, if -wN exists. If indeed nothing happens on these platforms, arguably the correct thing is to skip stripping entirely. Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
