Ghostscript

[fj]

If you want to see gs render the tiger, you need to change the default driver in Makefile.in as was previously done in the build instructions of ghostscript-8.15.4.

sed -i "s/bbox.dev$/x11.dev/" Makefile.in

Note that you can still see the bbox rendering by adding it to the command as follows:

bin/gs -sDEVCE=bbox -Ilib -dBATCH examples/tiger.eps

Also, using the command indicated to install the fonts will result in the following error:

tar: unrecognized option `--no-same-user'

Change '--no-same-user' to --no-same-owner'

[ken@…]

Taking this, since I've got gs-9 on my ToDo list.

[ken@…]

Looking at this (in gs 9.0.0) the current instructions in the book work for me - when running from rxvt-unicode, the tiger appears.

The no-same-owner change is covered by #3128.

[ken@…]

Seems to be a 32-bit-only issue. Randy told me he needed to specify --with-drivers=ALL to get the test to display the tiger. Found an old i686 installation, retested: no tiger. Reinstating the ticket until I've fixed it for i?86.

[Randy McMurchy]

Actually, ALL is the default. I made it into ALL,x11 and then it builds the x11 display driver. We don't need the sed, when configure parameters do the job.

[ken@…]

Well, I'm baffled about how the gs build system all fits together - I find the maze of .mak files impenetrable. If x11 is specifically built, the relevant objects are built rather earlier in the build, then also used in x11.so (instead of being built at the end, just before x11.so is linked). That is true for both i686 and x86_64, and I can see nothing different in the logs between the two arches (barring the arch name and an extra sizeof report from configure on 32-bit where it needs to find a 64-bit unsigned type, presumably for large files), so no idea why x86_64 doesn't need this.

Confirmed that x86_64 works ok with this (the 'tiger test'), so no harm in adding it for both, will do so.

[ken@…]

Fixed in r8716 and r8717

[bdubbs@…]

Milestone renamed