elwell@osu-eddie.UUCP (02/20/87)
First, allow me to apologize to rs@mirror for overstating my case :-). I misunderstood some of his complaints, and I must admit that recently I've been getting a little uptight about people complaining about TeX not doing "real" typesetting because it comes from a university. This is how I initially perceived the posting, and I'm afraid my noise/signal ratio was a little high. Sorry about that. By now I have accumulated a fair amount of experience figuring out how to make TeX "stand on its head and play Dixie" (usually successfully). With a little study/training, you can do amazing things (true for many systems, I admit). Lest I be accused of succumbing to the baby duck syndrome, let me now vent some of my complaints about TeX: 1. It's big and it's slow. Because of the way it was designed (machine independence at all costs) and its implementation as a monolithic Pascal program, it does a lot of unnecessary things. I've profiled it (an interesting experience in itself), and it spends most of its time reading from the input file, parsing it up, etc. The hyphenation and optimal line-breaking routines are actually respectably fast. 2. It's strictly batch-oriented. It can't do incremental reformatting (a big loss). 3. It represents everything as little rectangular boxes (rotate type? why would we want to do that?). I have the option of escaping into PostScript when I need to, but this means my document is then not really portable. I use TeX because (a) it is flexible, (b) it has an excellent paragraph builder, and (c) the price is right. I have for some time been thinking of implementing a system that uses many of TeX's algorithms, but implemented for high performance operation. Maybe someday I'll have the free time to do so. -- ==================== Total Nuclear Annihilation: Clayton Elwell The Ultimate Error Message. Elwell@Ohio-State.ARPA ...!cbosgd!osu-eddie!elwell ====================