ghostscript has a 300% size increase

Bug #255485 reported by Steve Langasek
2
Affects Status Importance Assigned to Milestone
ghostscript (Ubuntu)
Fix Released
High
Till Kamppeter
Intrepid
Fix Released
High
Till Kamppeter

Bug Description

Binary package hint: ghostscript

The latest ghostscript package which has just been merged into intrepid has a 300% size increase over the one in hardy, costing us 1.4MB of space on the alternate CDs that we didn't have to spare:

728K ubuntu/pool/main/g/ghostscript/ghostscript_8.61.dfsg.1-1ubuntu3_amd64.deb
2.2M ubuntu/pool/main/g/ghostscript/ghostscript_8.63.dfsg.1-0ubuntu1_amd64.deb

The primary cause of this appears to be that the newest upstream version of ghostscript is shipping a number of PS1 fonts in /usr/share/ghostscript/8.63/Resource/Font, which were not bundled before.

I'm pretty sure all of these fonts are available elsewhere in the archive, and probably on the install CD. They should be included by reference if possible (i.e., depend on the package that provides them), or if not, a way should be found to make these fonts optional for ghostscript.

Steve Langasek (vorlon)
Changed in ghostscript:
importance: Undecided → High
milestone: none → intrepid-alpha-4
status: New → Confirmed
Revision history for this message
Till Kamppeter (till-kamppeter) wrote :

I have asked the upstream developers on IRC and they tell that these fonts are the Ghostscript standard fonts and so they replace the gsfonts package.

So I would suggest then to take the gsfonts package from the distro to simplify maintenance. WDYT? Or was something special added to the gsfonts package?

Revision history for this message
Steve Langasek (vorlon) wrote :

Could/should the ghostscript binary package split the fonts out into a separate binary package, named "gsfonts"? There appear to be a number of packages that depend on gsfonts directly, so I think it makes sense to keep the binary name around. I suppose this would imply some compatibility layer, since the font filenames and install location have changed.

There's also defoma integration to be considered, since 'gsfonts' currently provides defoma support.

Otherwise, I don't see anything special in the gsfonts packaging; so if these fonts are the upstream replacement for gsfonts, I think we can make the switch, yes.

Revision history for this message
Martin Pitt (pitti) wrote :

Till, can we get this fixed on Monday or Tuesday, so that it makes alpha-4?

Changed in ghostscript:
assignee: nobody → till-kamppeter
Revision history for this message
Till Kamppeter (till-kamppeter) wrote :

Fixed in ghostscript 8.63.dfsg.1-0ubuntu2 by splitting the fonts into their own ghostscript-fonts sub package and made ghostscript depending on gsfonts OR ghostscript fonts. This solves the problem with the alpha4. The real solution is to eliminate the gsfonts package which should be done by the Debian maintainer at first (I have sent him mail already).

Changed in ghostscript:
status: Confirmed → Fix Released
Revision history for this message
Steve Langasek (vorlon) wrote :

Does using this OR dependency result in any integration issues in the short term? I.e., does ghostscript now expect the fonts to be present in a certain location, as provided by ghostscript-fonts?

Revision history for this message
Steve Langasek (vorlon) wrote :

btw, I don't think this OR dep really fixed anything - both ghostscript-fonts and gsfonts have ended up on the alpha-4 alternate CD, for instance. It ended up being less critical due to space savings elsewhere on the CD, but AFAICS this problem is still unresolved.

Changed in ghostscript:
milestone: intrepid-alpha-4 → intrepid-alpha-5
status: Fix Released → In Progress
Revision history for this message
Till Kamppeter (till-kamppeter) wrote :

There are two possibilities:

1. Total elimination of the new fonts in the Ghostscript package by using a modified .dfsg.2 source tarball. According to the GS developers the fonts in the Ubuntu gsfonts package are better (they plan to merge in the improvements of the Ubuntu package). The copyright notice in the font file can also iritate some users. The notice being AFPL is a bug, Ghostscript developers say that they are intended to be published under GPL.

2. Only replace the OR dependency with a pure gsfonts dependency.

Steve, WDYT?

Changed in ghostscript:
status: In Progress → Incomplete
Revision history for this message
Steve Langasek (vorlon) wrote :

examining the two packages, I can confirm that the fonts included in gsfonts are newer than the ones provided by ghostscript upstream. ghostscript upstream does, however, include two fonts that gsfonts does not (Dingbats and StandardSymL). So these two fonts should be shipped by ghostscript, whereas the others should continue to be provided by gsfonts for the time being.

I don't see any reason at all to modify the original ghostscript source, we just need to not ship the fonts in the binary package.

I suggest the following:
- recombine ghostscript-fonts into ghostscript proper (needs a replaces:, etc., for upgrades).
- depend only on gsfonts.
- strip the fonts provided by gsfonts out of the ghostscript binary package.
- (optionally?) provide symlinks mapping the gsfonts to the file names ghostscript looks for them under.

Changed in ghostscript:
status: Incomplete → Triaged
Revision history for this message
Steve Langasek (vorlon) wrote :

Here is a preliminary patch.

Do we need to also provide symlinks into /usr/share/fonts/type1/gsfonts, or will ghostscript manage to find these fonts on its own?

Revision history for this message
Till Kamppeter (till-kamppeter) wrote :

That is OK with me. Thanks for the patch.

Symlinks are not needed, GS 8.63 also works with the old font location and the old file names.

I will upload Ghostscript this way now.

Changed in ghostscript:
status: Triaged → In Progress
Revision history for this message
Till Kamppeter (till-kamppeter) wrote :

Steve, a new ghostscript package with your patch applied is uploaded now (0ubuntu4).

Changed in ghostscript:
status: In Progress → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.