The fundamental "correctness" of TeX/LaTeX is beyond question, as there are no alternatives for scientific work that comes even close in quality and performance (except for variants found on CTAN of course). In particular, your underlying assumption that a "modern application" is bound to be better is nonsense.
That's not entirely true. A lot of TeX/LaTeX is "correct", but the
fact that a TeX file can change the behaviour of the TeX parser is now widely considered to be bad design (in Knuth's defense, there were of course plenty of good reasons for this back when TeX got written). It is one
of the reasons why it is so hard to make a TeX-compatible system
without essentially being identical to TeX. Moreover, despite the
small size of the TeX source code, reimplementing it in something more
maintainable is a very tough job; the various "modules" are incredibly
tightly coupled together.
However, if you are a good LaTeX citizen, and don't attempt to write
complicated style files yourself, this of course hardly ever matters.
The main problem here is that making such a program without throwing
away at least some of the good features of TeX/LaTeX is hard, and the
market is small so there is not much manpower available. There is at
present not even a full-fledged GUI toolkit for mathematical typesetting (GtkMathView is good,
but this really should be integrated tightly with a text typesetting engine such as Pango).
One attempt you might want to have a look at is TeXmacs, which at least keeps a
very structured approach to document creation and has all the TeX
typesetting and layout algorithms.
"Standards" are not written as "do this, and everything else is nonstandard"
Which is exactly the problem with W3C standards. They are not. They are a collection of features that have surfaced in Netscape and IE, put together by people who do not understand that standards should be there to _reduce_ the number of ambiguous situations.
The reason why it is so damned hard to implement `all of the W3C standards' is that they do not follow the `keep it simple' strategy. Instead, there are rules, exceptions to the rules, slight modifications to the exceptions to the rules and so forth.
Unlike what most people think, W3C is just as bad as Netscape inventing , only this time they can call it an official standard. Open Source developers aren't even allowed to take part in the development of the W3C standards unless they have a couple of thousand dollars to spare. It is a _very_ closed process.
This is just my view as a developer of Mnemonic, but I'm sure Opera and Konqueror suffer from this problem as well.
Re:Maybe we'll see FreeBSD binaries again now :-)
on
Mozilla M12 Released
·
· Score: 1
Actually, the threads & dlopen bug in glibc was exposed by the Mnemonic team first;-) In general you're right though.
As one of the developers of Mnemonic, let me make a few remarks about the status of the project.
As with any project (including KDE's browser and Mozilla), Mnemonic has gone through several design stages and we have thrown away the entire codebase several times now. It's not very strange that only very few people know about it, since none of those times did we ever reach a point were the program did anything remotely useful.
At the moment, things are, however, progressing very rapidly. Apart from a very small core library (some 230 Kb), everything from network protocols to rendering engines is in separate modules. There is a rather decent generic XML parser, a completely GUI-toolkit independent rendering engine, network protocol modules (with SSL being added at this very moment) and a GTK based layer on top of that.
Just as the Konqueror team kept things relatively quiet, we have decided to work out the basics without giving too much publicity (although of course the code has always been available for anyone to look at). But as soon as the table layout algorithms have been debugged, we will start to release binaries (as that is probably the first time that the browser is really useful).
For more discussion, please join the mailing list; more info and status updates can be found at http://www.mnemonic.org.
Slashdot Mirror
The fundamental "correctness" of TeX/LaTeX is beyond question, as there are no alternatives for scientific work that comes even close in quality and performance (except for variants found on CTAN of course). In particular, your underlying assumption that a "modern application" is bound to be better is nonsense.
That's not entirely true. A lot of TeX/LaTeX is "correct", but the fact that a TeX file can change the behaviour of the TeX parser is now widely considered to be bad design (in Knuth's defense, there were of course plenty of good reasons for this back when TeX got written). It is one of the reasons why it is so hard to make a TeX-compatible system without essentially being identical to TeX. Moreover, despite the small size of the TeX source code, reimplementing it in something more maintainable is a very tough job; the various "modules" are incredibly tightly coupled together.
However, if you are a good LaTeX citizen, and don't attempt to write complicated style files yourself, this of course hardly ever matters.
The main problem here is that making such a program without throwing away at least some of the good features of TeX/LaTeX is hard, and the market is small so there is not much manpower available. There is at present not even a full-fledged GUI toolkit for mathematical typesetting (GtkMathView is good, but this really should be integrated tightly with a text typesetting engine such as Pango).
One attempt you might want to have a look at is TeXmacs, which at least keeps a very structured approach to document creation and has all the TeX typesetting and layout algorithms.
Which is exactly the problem with W3C standards. They are not. They are a collection of features that have surfaced in Netscape and IE, put together by people who do not understand that standards should be there to _reduce_ the number of ambiguous situations.
The reason why it is so damned hard to implement `all of the W3C standards' is that they do not follow the `keep it simple' strategy. Instead, there are rules, exceptions to the rules, slight modifications to the exceptions to the rules and so forth.
Unlike what most people think, W3C is just as bad as Netscape inventing , only this time they can call it an official standard. Open Source developers aren't even allowed to take part in the development of the W3C standards unless they have a couple of thousand dollars to spare. It is a _very_ closed process.
This is just my view as a developer of Mnemonic, but I'm sure Opera and Konqueror suffer from this problem as well.
Actually, the threads & dlopen bug in glibc was exposed by the Mnemonic team first ;-) In general you're right though.
Kasper
As with any project (including KDE's browser and Mozilla), Mnemonic has gone through several design stages and we have thrown away the entire codebase several times now. It's not very strange that only very few people know about it, since none of those times did we ever reach a point were the program did anything remotely useful.
At the moment, things are, however, progressing very rapidly. Apart from a very small core library (some 230 Kb), everything from network protocols to rendering engines is in separate modules. There is a rather decent generic XML parser, a completely GUI-toolkit independent rendering engine, network protocol modules (with SSL being added at this very moment) and a GTK based layer on top of that.
Just as the Konqueror team kept things relatively quiet, we have decided to work out the basics without giving too much publicity (although of course the code has always been available for anyone to look at). But as soon as the table layout algorithms have been debugged, we will start to release binaries (as that is probably the first time that the browser is really useful).
For more discussion, please join the mailing list; more info and status updates can be found at http://www.mnemonic.org.