piet@cs.ruu.nl (Piet van Oostrum) (09/22/89)
In article <426@laas.laas.fr>, ralph@nastassia (Ralph P. Sobek) writes: `In article <4861@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: `| From article <SPQR.89Sep14091859@hilliard.ecs.soton.ac.uk>, by spqr@ecs.soton.ac.uk (Sebastian Rahtz): `| " `| " Is there a newer version of a DVI to Postscript converter out there `| " anywhere? `| `| Yes. A version of dvi2ps by Piet van Oostrum has some useful `| new features. I will have yet another version soon -- I'm working `| out what I hope are the last few bugs. `| `Oh my god!! Another version?!! I hope that there is someone to `coordinate all these versions. We alone have 3 versions of dvi2ps. `Normally the most recent has 2.11 as version number. ` `It's a shame that Piet van Oostrum did not add his interesting `modifications to the latest dvi2ps! For example, dvi2ps already `accepts PK and GF font files and in seperate directories. ` `Is someone (Piet?) willing to create dvi2ps 2.12? Maybe, it already exists. ` Let me first tell, what I know about dvi2ps genealogy and how I got involved. I started working on dvi2ps in May 1986, so my version has been around for a resonable time. When we started using TeX, I encountered a couple of problems: 1. We were using a VAX11/750 with Unix BSD 4.1. Our system manager didn't want to switch to 4.2, and the dvi2ps I found on the Unix tape didn't work on 4.1, so I had to patch the program anyway. 2. Our disk was almost full and couldn't take all of the PXL files, which was the only font format supproted then by the supplied dvi2ps. So I decided to add support for PK files. I think it is silly to use PXL files, anyway. I just coded the algorithms from pktype into dvi2ps. 3. I found it very annoying that the LaserWriter run out of Virtual Memory after 30-40 pages (sometimes even after a single page, if you had many fonts. I think I really got upset when I wanted to print fontsamples of the supplied fonts and couldn't get a page with every line a different font. So I added code that estimated how much memory was left in the LW, and if a certain threshold was reached, no more characters were downloaded, but they were printed on an individual basis. This has been very satisfactory because we never run out of VM anymore. I also looked very carefully in the supplied Postcript code and found a number of places where extra save/restore pairs could decrease the amount of VM used. These were the basic initial changes I made (along with a number of minor cleanups). Later I added code to use Postscript fonts. I got this code from somebody else, integrated it, cleaned it up and made it easily parametrizable. The other changes were just useful options and bug changes. I have advertised the availability of my version on TeXhax, and a more recent version on comp.text. Several people in Europe, USA and Australia have received copies. My own version is now 2.48, and I have a few bug patches waiting. Apparently, other people were also working on different modifications of dvi2ps. The versions of dvi2ps that float around, form a Directed Acyclic Graph (actually, I'm not sure that it is acyclic :=). The basic problem, I think, was that there was no means of finding out who did what. TeXhax was the only way I had to find out what was going on. There has been a discussion on merging the various versions of dvi2ps on Texhax, and once it seemed that somebody was going to make a single version with all the enhancements in it, but apparently that failed (I got this info from Pierre MacKay) I am not going to defend the fact that I have maintained a separate version of dvi2ps. In fact when I saw the announcement of dvi2ps 4.0 by Tomas G. Rokicki, I decided that it would be time to discontinue development of my version. I have not yet seen his version (I can't FTP yet, so I have to find it some other way). If it does have everything I need (and I suspect it does) I will probably switch to his version. I will put the final bug patches in my version, and I will spuuly it to anyone who wants it, but I suggest that we agree on a single supported standard version of dvi2ps. There seems to be also a version called dvitps by Stephan von Bechtolsheim, but I don't know anything about it. Maybe these two gentlemen could even agree on merging their programs? From what I have read these persons have done a very good job. The best way to standardize in my opinion is to keep the standard version up to date on the Unix tape, and to distribute patches on Usenet or through a mailing list. I am willing to give my code, suggestions, etc to get a good standard version, but I'm not going to reinvent the wheel. That is the primary reason for me to stop enhancing dvi2ps 2.48. I have cross-posted this to comp.text to get a greater audience, but followups only to comp.lang.postscript. Should I also send something to teXhax for those poor people without net.access? -- Piet van Oostrum, Dept of Computer Science, University of Utrecht Padualaan 14, P.O. Box 80.089, 3508 TB Utrecht, The Netherlands. Telephone: +31-30-531806 Internet: piet@cs.ruu.nl Telefax: +31-30-513791 Uucp: uunet!mcvax!hp4nl!ruuinf!piet
kevin@math.lsa.umich.edu (Kevin Coombes) (09/22/89)
Okay, here goes. I've spent the last few weeks working on merging several versions of dvi2ps---namely, the three versions available at the june.cs.washington.edu archive, by Larry Denenberg, Peter Damron, and Jing-bai Wang. I also have Van Jacobson's version. I have a copy of Tomas Rokicki's version 4.0 of dvips, but haven't yet had a chance to look at it. The merge isn't yet complete, nor is it ready for general consumption. Right now, it can use PK, GF, or PXL files (with the available choices set at compile time), allows flexible organization of font directories and supports environment variables to help with this. It searches for fonts in a reasonable manner, and will use postscript to rescale fonts if it can't find them in the right size. It supports various specials, including psfig. The point is: I'm willing to invest some more time in producing version 3.0 (or whatever) of dvi2ps. I can see from this discussion that there is some interest. What I'd like (by e-mail, and I'll post summaries) is information on what other versions are available, how to get them, and what features people most want to see. I have heard of the following versions, but have not seen copies: dvitps by Stephan von Bechtolsheim dvitops by James Clark dvi2ps by Piet van Oostrum. If anyone is interested in helping to produce a merged version of dvi2ps, or in testing such a version if/when it becomes available, I'd also appreciate hearing that. Kevin Coombes <kevin@math.lsa.umich.edu> <Kevin_Coombes@ub.cc.umich.edu>