Broken dot-new config handler
by Xsane from LinuxQuestions.org on (#4WVZE)
I posted this in 'Requests for -current', but it didn't get any traction; so lets
try a separate thread. Any comments?
The dot-new handler falls down when the configuration file has a version in its
path. For example, every upgrade of the ghostscript package is leaving cruft due
to cidfmap.new
Code:/usr/share/ghostscript/
a"a"a" 9.22
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmap
a"a"a" 9.23
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmap
a"a"a" 9.24
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmap
a"a"a" 9.25
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmap
a"a"a" 9.26
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmapThe only useful instance for cidfmap.new would be a rebuild
(same version; same path).
Any sys-admin using a custom cidfmap must be used to reinstalling it
on every upgrade; so having to do so on a rebuild should not be a
surprise. Meaning, that IMO cidfmap should not be using the dot-new
handler. I suppose the handler could be rewritten for this corner
case, but it probably wouldn't be a good ROI.


try a separate thread. Any comments?
The dot-new handler falls down when the configuration file has a version in its
path. For example, every upgrade of the ghostscript package is leaving cruft due
to cidfmap.new
Code:/usr/share/ghostscript/
a"a"a" 9.22
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmap
a"a"a" 9.23
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmap
a"a"a" 9.24
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmap
a"a"a" 9.25
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmap
a"a"a" 9.26
a"** a""a"a" Resource
a"** a""a"a" Init
a"** a""a"a" cidfmapThe only useful instance for cidfmap.new would be a rebuild
(same version; same path).
Any sys-admin using a custom cidfmap must be used to reinstalling it
on every upgrade; so having to do so on a rebuild should not be a
surprise. Meaning, that IMO cidfmap should not be using the dot-new
handler. I suppose the handler could be rewritten for this corner
case, but it probably wouldn't be a good ROI.