We are using Epic 5.1C on Sun Solaris to generate PostScript files from XML source using "pubformat" in a C-shell script. The XML documents are all UTF-8 encoded. This works fine with English and all European languages. We'd now like to also process C/J/K output in this environment and would like to know what we need to do in our Sun Solaris environment to have the C/J/K fonts available. My first thought was that since we're using and specifying UTF-8, they would just be there. Interestingly, if we bring up a Japanese UTF-8 document in the Editor, it displays fine. It even displays fine when we do a print preview. But, when generating the PostScript, the fonts aren't found. We don't have native C/J/K installs of SunOS, so maybe that's required? We run our environment in background batch jobs, so logging into a native language environment is not an option.
Has anyone already done what we'd like to do or have any suggestions on what we need to do? Is there any detailed Arbortext documentation that we don't know about?
I hardly know anything about Unix, but when you generate your PostScript, do you use a printer driver? If so, does this printer have those fonts? If you don't use a printer driver, does whatever you *do* use know about these fonts?
We use the Arbortext "pubps" script, which doesn't use a printer driver. But, that's a good question. How does pubps know where to get the fonts? It seems to know where to find characters encoded as UTF-8, so shouldn't it be able to find any UTF-8 character?
Arbortext Help 489, 5817, 7592, and 13019 cover pieces of this issue. Note the set command mentioned in 7592. The locale discussion, however, probably applies more to someone authoring with Epic Editor. For batch composition, there must be a way to compose the UTF-8 output if the fonts are already available on Solaris.
When I added support for ArialCE in our Solaris environment, I used a technical note supplied by Arbortext TPS or Consulting entitled "Preparing and Using a Custom Print Font for ADEPT on Unix". It worked in 2000 and again in 2004 using Epic 4.3.1 and earlier tools.
There's some possibility that Epic 4.4 doesn't usually require manual edits to charent.cf. However, the CJK support appears new and therefore may differ.
I expect to be visiting the CJK issue sometime in the future. I'd be interested to hear how others have done it.