rs@mirror.UUCP (02/13/87)
>So, anyway, why not TeX for ``real'' typesetting? My personal opinion is >that it's been ignored by the industry because it's *free*, or nearly. >My newspaper paid around *$10,000* for the rotten software it has now. >TeX costs about $100. Maybe $150. Have you ever tried to get TeX to set newspaper type? I have, and it's an enourmous waste of time and computer power. Face it, there is no way to avoid lots of hyphenation when setting 9pt type in a 23pica (~2in) column. Second, I don't think TeX is a good fit for adwork or custom typehouses. One major strength (and, in the hands of incompetents, the biggest weakness) of typehouses and design firms is their huge font library. TeX can use any fonts for which it has the right information, but I challenge anyone to calculate bounding boxes for a "shadow" typeface (i.e., outlined letters with deep shadows) where kerning is totally dependant on the adjacent letters -- how deep the shadows can cut into each other. As far as bookwork goes, TeX's output model is not powerful enough. Modern systems (e.g., Xyvision, Bedford, Texet, etc.) all have the ability to back-up and re-set the previous page using a different format if final page leaves you with a widow or such. TeX never looks back -- once the page is set, it's forgotten about. TeX also doesn't have the concept of "exception pages" where you want to override your style rules for this one page because it has, say, five pictures on it. Before you point out Addison-Wesley's use, take a good look at the design of the books they use, and compare it with something like a technical manual with lots of diagrams. Fourth, it's just too hard to do too many things in TeX. You don't need to use \vbox to 3.5i{...} to get vertical justfication. Heck, the newspaper composition system I bought five years ago had vertical justification, and the syntax was very similar, including explicit marking of "spread points." This can be corrected with a good macro package, but who's going to write and maintain it? Finally, the cost of $100 for TeX is a fictitious one. You still have to buy a machine, terminals, train your users to use the machine's native OS and editor, etc. You have to support them, and you have to deal with ongoing maintenance. There are no firms out there that will install, write macros, and keep you supplied with maintenance releases (Knuth just found a TeX bug last month, so it isn't bug-free yet...). Try telling any large typehouse, let alone any large newspaper, that you've got a great new composition system for them that's cheap; the only thing missing is software support from the vendor. Except for the output model, I admit that none of these problems are insurmountable. Still and all, that may be good enough to do many of the commercial typesetting jobs now being done more traditionally. I just think that the TeX evangelists are over-eager and over-optimistic: TeX has its places, but the commercial market is not one of them, yet. -- Rich $alz "Drug tests p**s me off" Mirror Systems, Cambridge Massachusetts rs@mirror.TMC.COM {adelie, mit-eddie, ihnp4, harvard!wjh12, cca, cbosgd, seismo}!mirror!rs
elwell@osu-eddie.UUCP (02/17/87)
I think it's time to toss in a few facts to give the disinformation some flavor... In article <212300001@mirror> rs@mirror.UUCP writes: >Have you ever tried to get TeX to set newspaper type? I have, and it's >an enourmous waste of time and computer power. Face it, there is no >way to avoid lots of hyphenation when setting 9pt type in a 23pica (~2in) >column. Well, let's see. 4 columns per page (~1.5 in wide columns), in 8pt Helvetica Narrow. Does that count? TeX has an extremely good hyphenation algorithm. It tries not to hyphenate, to be sure, but if density is more important than readability, just lower the hyphenation penalty (\hyphenpenalty=mumble). >Second, I don't think TeX is a good fit for adwork or custom typehouses. >One major strength (and, in the hands of incompetents, the biggest weakness) >of typehouses and design firms is their huge font library. TeX can use >any fonts for which it has the right information, but I challenge anyone >to calculate bounding boxes for a "shadow" typeface (i.e., outlined >letters with deep shadows) where kerning is totally dependant on the >adjacent letters -- how deep the shadows can cut into each other. One of TeX's advantages is that it DOESN'T use bounding boxes. It uses arbitrary pair-based kerning and ligatures (that interact correctly). If you can kern it manually, you can tell TeX how to do it once and never have to worry about it again. I challenge any other typesetting system to take a Russian phrase in standard Roman orthography and translate it into the appropriate Cyrillic characters completely automatically when the phrase is set in a Cyrillic font (heh heh). >As far as bookwork goes, TeX's output model is not powerful enough. >Modern systems (e.g., Xyvision, Bedford, Texet, etc.) all have the >ability to back-up and re-set the previous page using a different >format if final page leaves you with a widow or such. TeX never >looks back -- once the page is set, it's forgotten about. TeX >also doesn't have the concept of "exception pages" where you want >to override your style rules for this one page because it has, say, >five pictures on it. Before you point out Addison-Wesley's use, >take a good look at the design of the books they use, and compare >it with something like a technical manual with lots of diagrams. Aside from the fact that I've done technical manuals with lots of diagrams with no problem, this is just wrong. TeX will reset a page as many times as you want (although it does have a [high] threshold to catch endless loops in the output stage). For example, the TeXbook (the author's combined tutorial and reference manual) has an example of an output macro that will repeatedly reset the page until it gets two columns balanced to the same number of lines (great for parallel translations). As for exception pages, I'm suprised to hear that, since I've done it myself with no difficulty ("well, this page is a chapter heading, so we'll put the page number at the bottom, clear the page header, etc."). To quote a bad TV commercial, "Where's the beef?" >Fourth, it's just too hard to do too many things in TeX. You don't need >to use \vbox to 3.5i{...} to get vertical justfication. Heck, the >newspaper composition system I bought five years ago had vertical >justification, and the syntax was very similar, including explicit >marking of "spread points." This can be corrected with a good macro >package, but who's going to write and maintain it? I don't even understand the question. Maybe I haven't had enough practice doing bad layout, but my columns & pagees come out vertically justified OK. >Finally, the cost of $100 for TeX is a fictitious one. You still have >to buy a machine, terminals, train your users to use the machine's >native OS and editor, etc. You have to support them, and you have to >deal with ongoing maintenance. There are no firms out there that will >install, write macros, and keep you supplied with maintenance releases >(Knuth just found a TeX bug last month, so it isn't bug-free yet...). >Try telling any large typehouse, let alone any large newspaper, that >you've got a great new composition system for them that's cheap; the >only thing missing is software support from the vendor. Poor little baby, all alone in the big bad marketplace... [1/2 :-)] You are, of course, ignoring the worldwide and highly active TeX user's group. I won't even mention the fact that several top-notch typography houses here in Columbus are *real* interested in TeX because it can do things nothing else will. How many vendors of turnkey (read: proprietary) systems are interested in giving you real support? They've already got you by the privates, and they don't need to care anymore. >Except for the output model, I admit that none of these problems are >insurmountable. Still and all, that may be good enough to do many of >the commercial typesetting jobs now being done more traditionally. >I just think that the TeX evangelists are over-eager and over-optimistic: >TeX has its places, but the commercial market is not one of them, yet. Since the output model is not the limiting factor (although, being amazingly flexible, it does take some study to understand), I don't see your point. In my experience, TeX and TeX output have survived quite well in the commercial market, thank you. >Rich $alz "Drug tests p**s me off" >Mirror Systems, Cambridge Massachusetts rs@mirror.TMC.COM >{adelie, mit-eddie, ihnp4, harvard!wjh12, cca, cbosgd, seismo}!mirror!rs -- ==================== Total Nuclear Annihilation: Clayton Elwell The Ultimate Error Message. Elwell@Ohio-State.ARPA ...!cbosgd!osu-eddie!elwell ====================
rs@mirror.UUCP (02/18/87)
Sigh; apparently I wasn't clear enough. TeX can set newspaper type, of course. It's just a gross waste of cycles, that can more effectively used elsewhere. Quoting the TeX source, "the line-breaking problem can be regarded as a special case of the problem of computing the shortest pah in an acyclic network." I said TeX couldn't do things like properly kern shadow typefaces. I didn't mean to imply that other software could: just that in custom typework (in particular, custom adwork) lots of stuff is still done by hand because it has to be. It doesn't matter if you call the data bounding boxes, or kerning pairs: if you need to get in between the letters with an X-Acto knife, a computer solution will not do it. >I challenge any other typesetting system to >take a Russian phrase in standard Roman orthography and translate it into >the appropriate Cyrillic characters completely automatically when the phrase >is set in a Cyrillic font (heh heh). I don't know enough about Russian to fully reply, but most modern composition systems have very flexible ways of mapping the input (working) character set to the output character set on a per-font basis. They have to have that capability in order to properly generate output for the back-end typesetters. I also said, >As far as bookwork goes, TeX's output model is not powerful enough. >Modern systems (e.g., Xyvision, Bedford, Texet, etc.) all have the >ability to back-up and re-set the previous page using a different >format if final page leaves you with a widow or such. And saw this response: >... this is just wrong. TeX will reset a page as many times >as you want (although it does have a [high] threshold to catch endless >loops in the output stage). Apparently I wasn't clear enough. Suppose I am composing page "n" of my book, and I find it's the last page of my chapter, and I just cannot set it properly. I want to go back to page "n-1" and rip it up so that, e.g., instead of doing two-column format I do one-column format. TeX cannot do this. TeX's model is a linear flow through the composition process: linebreak, paragraph, vertical justification, output. There is no backward flow, to try out, e.g., a different page layout with different linelengths and such. Certainly not going back to previous pages. Technically, a chapter beginning or a chapter ending are not exception pages. In psuedo-TeX, an exception page happens when I change my style guide for one page becuase the collective badness is too great using all the regular styles. It should happen automatically, when the system is unable to set things to meet my criteria. I also said: >Finally, the cost of $100 for TeX is a fictitious one. You still have >to buy a machine, terminals, train your users to use the machine's >native OS and editor, etc. You have to support them, and you have to >deal with ongoing maintenance. There are no firms out there that will >install, write macros, and keep you supplied with maintenance releases >(Knuth just found a TeX bug last month, so it isn't bug-free yet...). >Try telling any large typehouse, let alone any large newspaper, that >you've got a great new composition system for them that's cheap; the >only thing missing is software support from the vendor. And saw this response: >Poor little baby, all alone in the big bad marketplace... [1/2 :-)] You >are, of course, ignoring the worldwide and highly active TeX user's group. I'm a member of TUG, and am on the TexHax mailing list. I'm well aware of the user's group, and the types of questions they get, and problems they answer, and work they're doing. Continuing: >How many vendors of turnkey (read: proprietary) systems are interested >in giving you real support? They've already got you by the privates, and >they don't need to care anymore. Oh come on. You want to make a blanket condemnation of every vendor in the type business, that's your right. Buying a PC from FlyByNite, Inc., exactly the same thing as buying a million-dollar text-and-graphics system from, Computer Composition (to pick a name). Large type-houses, and especially large newspapers, need systems with promised response time and promised down time. These things are usually written into their RFP. At worst case, the client has legal recourse. Let's get off this silly side issue, OK? I'll be happy to response to E-mail on this topic. -- Rich $alz "Drug tests p**s me off" Mirror Systems, Cambridge Massachusetts rs@mirror.TMC.COM {adelie, mit-eddie, ihnp4, harvard!wjh12, cca, cbosgd, seismo}!mirror!rs