'texconfig' modifies /usr/share/texmf/dvips/config/config.ps but
'dvips', and 'xdvi', finds via 'kpathsea' library another config
file in /usr/share/texmf/dvips/base/config.ps . The net effect
is that configuration changes are not effective, apparently "random"
bitmap files are generated and printing fails as no print command
is defined. Also, if printing to a file, an output has a wrong
resolution and is in A4 format even if 'letter' was requested.
To fix a file base/config.ps should be renamed, or simply removed, or
made into a link to config/config.ps. Finding that and repairing will
be highly non-trivial for people without a close knowledge of a layout
of a TeX installation and 'kpathsea'.
Yes. This is a regression from 1.0.7-7. Thanks for the report.
This is caused by the dvipsk586.patch in dvipsk-jpatch-p1.5b.tar.gz.
Matt, do you know why this patch installs base/config.ps? Who made this
patch, so that I can ask them?
ASCII in Japan. I have no idea who to talk to.
I would strongly suspect that this is an overenthusiastic patching. Old
versions of tetex did use dvips/base/config.ps for a configuration file
and it is likely coming from there. At the first look this seems to be
a regular "sample" config.ps and nothing more.
'dvips' will look for 'config.ps' anywhere in that tree (look into
web2c/texmf.cnf) and that is the problem because it is finding the wrong
one; "wrong" in that sense that 'texconfig' writes specifically to
This defect considered MUST-FIX for Florence Gold
How does it help ASCII in Japan? I don't see it..
I'm edging towards just reverting that particular change.
Ah, there is a newer version of the dvipsk-jpatch patchset (1.5d), which fixes
Fixed in tetex-1.0.7-14.
I have changed things in this area to finally get Japanese dvips working.
Now there is 'dvips' and 'pdvips', and texmf.cnf has different settings for
This bug _should_ still be closed, but I'm reopening it to make sure, until