host_os_name value in config-scripts has changed to 'solaris'

[l1gi]

Variable host_os_name value has changed on Solaris from sunos to solaris some time ago. This change updates config-scripts to keep values used for Solaris 5.10 (sunos), but also sets relevant options on Solaris 5.11 (solaris). This change decreases the number of configure options necessary to build CUPS on current Solaris.

Please consider merge.

Thank you,
m.

[l1gi]

Could I kindly ask you to review this Solaris specific change, please?

Thank you,
m.

[LumioseSil]

Make sure to run autoconf --force @l1gi

[michaelrsweet]

@AtariDreams NEVER run "aclocal" on the CUPS repository - it will damage the config.h.in file and render the local sources unbuildable. We do not use automake or aclocal, just autoconf (autoconf -f) to rebuild the configure script.

[michaelrsweet]

@l1gi I've been in standards meetings all week, but just FYI it is not reasonable to expect code review after 6 hours.

[michaelrsweet]

Please respond to my question about /system. Also, you missed the sunos reference in the cups-threads.m4 file.

[michaelrsweet]

Not having used a recent version of Solaris, this looks like something that needs more checking - certainly the last version of Solaris that CUPS actually supported did not have a /system hierarchy, so I'd expect there to be a directory existence check.

[l1gi]

True, will update this.

[l1gi]

Hi Michael,

thanks for looking into this. Your objections caused I did more testing on how that behaves on Solaris 10 and I must say I was wrong. Even on Solaris 10 (which is the last Oracle supports) the host_os_name is already solaris. So the question is how sunos got there and whether we should keep it or replace it by solaris. What do you think (or remember)? :)

Solaris 11 builds with or without -D_POSIX_PTHREAD_SEMANTICS, but Solaris 10 needs this. I will take care of this one also.

Thanks for any advice,
m.

[michaelrsweet]

WRT what host_os_name contains, that is controlled by the GNU config.sub/guess scripts which have been updated several times over the years. And prior to us using that we just use the uname output, which probably still reports SunOS on Solaris (but I really don't have a way to test that...)

Anyways, checking the config.sub/guess scripts, it looks like they still report sunos on platforms that are running pre-Solaris builds (if you'd even be able to get working hardware for that!) so for now let's use both sunos and solaris in the configure scripts.

[l1gi]

WRT what host_os_name contains, that is controlled by the GNU config.sub/guess scripts which have been updated several times over the years. And prior to us using that we just use the uname output, which probably still reports SunOS on Solaris (but I really don't have a way to test that...)

That's true of course:

$ uname -a
SunOS ulx-0 5.11 11.4.60.147.0 i86pc i386 i86pc non-virtualized

Anyways, checking the config.sub/guess scripts, it looks like they still report sunos on platforms that are running pre-Solaris builds (if you'd even be able to get working hardware for that!) so for now let's use both sunos and solaris in the configure scripts.

Okay, so I have just added solaris branches where necessary leaving sunos as-is and made a directory check for CUPS_STATEDIR.

[LumioseSil]

@michaelrsweet Thoughts?

[michaelrsweet]

Still needs a few tweaks...

[michaelrsweet]

Need an 'else' clause for this.

[michaelrsweet]

Needs to look for both sunos and solaris here, something like:

AS_IF([test $host_os_name = sunos -o $host_os_name = solaris], [

[l1gi]

Let me know if it's okay now.

Thank you,
Martin

[michaelrsweet]

Looks good now, thank you!