Mailing List Archive

On Sun, 2002-12-22 at 12:32, Tomasz Wegrzanowski wrote:
> texvc is done, testing found no bugs.
>
> Please install it on all production Wikipedias.

Can you put it into CVS or send a diff?

I don't want to just copy the test.wikipedia.org files since that's got
other not-fully-tested code in it.

-- brion vibber (brion @ pobox.com)

On Sun, Dec 22, 2002 at 12:56:26PM -0800, Brion Vibber wrote:
> On Sun, 2002-12-22 at 12:32, Tomasz Wegrzanowski wrote:
> > texvc is done, testing found no bugs.
> >
> > Please install it on all production Wikipedias.
>
> Can you put it into CVS or send a diff?
>
> I don't want to just copy the test.wikipedia.org files since that's got
> other not-fully-tested code in it.

Here it is. Patch against most recent CVS.

--- Tomasz Wegrzanowski <taw@users.sourceforge.net> wrote:
> texvc is done, testing found no bugs.
>
> Please install it on all production Wikipedias.

Tomasz, could you please also implement a <tex>...</tex> which goes
straight to TeX, so that matrices, commutative diagrams, chess
diagrams, chemical structure formulas, music notes and flow charts can
be typeset conveniently? Or would you prefer that I do it?

Has the issue of non-specific error messages been resolved?

Axel

__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

On Mon, Dec 23, 2002 at 05:06:39PM -0800, Axel Boldt wrote:
> --- Tomasz Wegrzanowski <taw@users.sourceforge.net> wrote:
> > texvc is done, testing found no bugs.
> >
> > Please install it on all production Wikipedias.
>
> Tomasz, could you please also implement a <tex>...</tex> which goes
> straight to TeX, so that matrices, commutative diagrams, chess
> diagrams, chemical structure formulas, music notes and flow charts can
> be typeset conveniently? Or would you prefer that I do it?

No. I'm not going to implement that and I'm strongly against such
feature. http://test.wikipedia.org/wiki/Why_we_can't_just_use_LaTeX

texvc supports matrices.

> Has the issue of non-specific error messages been resolved?

Yes.

> No. I'm not going to implement that and I'm strongly against such
> feature. http://test.wikipedia.org/wiki/Why_we_can't_just_use_LaTeX

I tend to agree here, let's stick with the current solution until we find
a problem we cannot solve otherwise. This addition was primarily intended
for mathematics, and while other creative uses are possible, I think this
leads us into dangerous territory. Do we really *want* people to draw
diagrams etc. in LaTeX? Or shouldn't they just upload them as images? Many
people in the field of Maths are familiar with TeX, but for most others,
it's just gibberish, and I'm afraid people might start using it as a
replacement for things that are currently not supported by our wiki-
syntax. One of these things is a powerful table and image layout syntax.

I think the attack strategy argument is especially strong. You can see
texvc as a kind of firewall for code, we can adjust its settings, but it's
good to have it.

Regards,

Erik

On Mon, Dec 23, 2002 at 02:21:24AM +0100, Tomasz Wegrzanowski wrote:
> On Sun, Dec 22, 2002 at 12:56:26PM -0800, Brion Vibber wrote:
> > On Sun, 2002-12-22 at 12:32, Tomasz Wegrzanowski wrote:
> > > texvc is done, testing found no bugs.
> > >
> > > Please install it on all production Wikipedias.
> >
> > Can you put it into CVS or send a diff?
> >
> > I don't want to just copy the test.wikipedia.org files since that's got
> > other not-fully-tested code in it.
>
> Here it is. Patch against most recent CVS.

It seems it hasn't been commited to CVS and put on production Wikipedias yet.

Brion, are there any problems with this diff ?

I'd really like to have <math> available as soon as possible.

On mer, 2002-12-25 at 13:36, Tomasz Wegrzanowski wrote:
> It seems it hasn't been commited to CVS and put on production Wikipedias yet.
>
> Brion, are there any problems with this diff ?

I haven't had a chance to look at it -- it's Winter Economic Frenzy
Holiday and I'm a little busy. :)

> I'd really like to have <math> available as soon as possible.

I'll try and get it set up today or tomorrow.

-- brion vibber (brion @ pobox.com)

On dim, 2002-12-22 at 17:21, Tomasz Wegrzanowski wrote:
> Here it is. Patch against most recent CVS.

I've added it to CVS, updated the install docs, added patch-math.sql to
the patch-list.txt, and put test.wikipedia.org on the new CVS version
(plus a slight patch for the InnoDB tables; taw and magnus, I put
innodb.diff into the source dir there for reference in case you change
the running stuff again.)

Quick question before I go installing this: if I set "Leave it as TeX
(for text browsers)", I get HTML-rendered tables and things, not raw
TeX. Is it doing the wrong thing, or is the option mislabeled?

-- brion vibber (brion @ pobox.com)

On Thu, Dec 26, 2002 at 01:41:58PM -0800, Brion Vibber wrote:
> On dim, 2002-12-22 at 17:21, Tomasz Wegrzanowski wrote:
> > Here it is. Patch against most recent CVS.
>
> I've added it to CVS, updated the install docs, added patch-math.sql to
> the patch-list.txt, and put test.wikipedia.org on the new CVS version
> (plus a slight patch for the InnoDB tables; taw and magnus, I put
> innodb.diff into the source dir there for reference in case you change
> the running stuff again.)
>
> Quick question before I go installing this: if I set "Leave it as TeX
> (for text browsers)", I get HTML-rendered tables and things, not raw
> TeX. Is it doing the wrong thing, or is the option mislabeled?

There were 3 options first (PNG, HTML, TeX), and 4 later (PNG,
conservative HTML, liberal HTML, TeX).
Are you sure you have right version of Language*.php ?
If it's too old label TeX would mean liberal HTML.

Tomasz Wegrzanowski wrote:

>Are you sure you have right version of Language*.php ?
>
I might accidentially have overwritten Language.php at the test site
while working on the improved Recent Changes. Sorry.

Magnus

Tomasz Wegrzanowski wrote:

>texvc is done, testing found no bugs.
>Please install it on all production Wikipedias.

*Please* don't.

texvc is full of problems; here are the two primary ones:

1: A simple equation like <math>x + y = z</math> looks ugly
(uglier than ordinary wiki markup with or witout italics)
under any user preference -- and simply wrong under the default.

This is primarily because Tomasz has neglected
to implement TeX's spacing algorithm
in the HTML version of the texvc rendering.
Sure, he could now go and programme it.
But why bother when Don Knuth already did?
Which brings us to primary problem 2:

2: The entire philosophy of texvc is wrong from start to finish.


I intended to write an essay on texvc and how it *should* work
on MetaWikipedia -- where such an essay belongs --
once I got back online (today or later in the weekend),
although I didn't know (until reading my email just now)
that I would have this impending deadline to delay its implementation.
I'll try to get it done today (not just sometime this weekend).

In the meantime, I think that installing texvc
wouldn't simply fail to solve our math formatting problems
but would actually make things worse.
Math formatting isn't a pressing issue,
but in the end we will certainly want it.
We should do it right. texvc is not right.
And it's so much off from right, in fact,
that it will be harder to make things right later
if people start using it now. (Details in my essay.)


-- Toby

Toby Bartels is a troll and I'm not going to listen any more to what he says.

--- Erik Moeller <erik_moeller@gmx.de> wrote:
> > No. I'm not going to implement that and I'm strongly against such
> > feature. http://test.wikipedia.org/wiki/Why_we_can't_just_use_LaTeX
>
> I tend to agree here, let's stick with the current solution until we
> find a problem we cannot solve otherwise. This addition was primarily
> intended for mathematics,

Well, if so, it might have been a good idea to involve more of the
people who are actually active in Wikipedia's math area. Right now,
Toby and I have both stated that we are not too fond of the texvc
approach.

> Do we really *want* people to draw
> diagrams etc. in LaTeX? Or shouldn't they just upload them as images?

Or why not upload formulas as images, and be done with the whole
debate?

No, people should definitely be able to create math formulas, chess
diagrams, chemical structure diagrams, music scores and flow charts in
LaTeX, because that's the wiki way: these graphics can then be modified
by directly editing the source code, while uploaded images can only be
redrawn from scratch or pixel edited. Granted, the TeX syntax is often
obscure, but change a little bit here, preview, change a bit there,
preview, and you figure it out. And unlike texvc, it's widely used and
there are tutorials all over the place.

Arguably, the GFDL requires that if I create a graphic with some
scriptable software such as gnuplot, latex, ChemSketch or gimp, I am
required to make available the modifiable source code used to produce
the picture. Legalities aside, it is always a good idea to do so; what
better way than to simply upload *only* the source code and let
Wikipedia deal with the compilation and presentation.

> it's just gibberish, and I'm afraid people might start using it as a
> replacement for things that are currently not supported by our wiki-
> syntax. One of these things is a powerful table and image layout
> syntax.

Image layout and table support is currently available on Wikipedia with
HTML syntax, simpler and more powerful than TeX's gibberish table
support. I don't see how this makes an argument against TeX. What other
examples did you have in mind?

Axel

__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

> Well, if so, it might have been a good idea to involve more of the
> people who are actually active in Wikipedia's math area. Right now,
> Toby and I have both stated that we are not too fond of the texvc
> approach.

I can understand that - after all, you are familiar with TeX and you want
to use its maximum power. But from a Wikipedia-wide perspective, we have
to keep other factors in mind.

> No, people should definitely be able to create math formulas, chess
> diagrams, chemical structure diagrams, music scores and flow charts in
> LaTeX, because that's the wiki way: these graphics can then be modified
> by directly editing the source code

I know LaTeX and I use it for all my written correspondence. That means
that I also know how unreadable it can get, even if you have the 500 page
Kopka introduction or something similar near your desk. It's possible to
do great things with LaTeX, but then, why not have a Python backend with
some drawing library, or maybe a POV-Ray backend as well ..

Instead of arbitrarily providing such functionality, we should, for all
different types of problems, think carefully what the best tool for the
job is. In Maths, TeX/LaTeX is very popular, so it may well be the best
tool for the job here. But for music, maybe the recently mentioned GNU
Lilypond would be better and easier to learn/use? Basic diagrams might be
best handled in SVG or something like that, for plots a gnuplot backend
might be nice. And so forth, and so on. Aside from the usability
advantage, you also get the geek factor of being able to play with many
different toys.

If we do not do this, we force people who want to participate in the wiki
process to learn tools that may be suboptimal, even though there may be
better and more popular tools for that specific job. From a usability
perspective, that's a bad idea. By limiting initially the scope of TeX
use, we avoid this usability trap. If it turns out that the music people
prefer TeX to Lilypond, we'll send some hired goons to Tomasz and persuade
him to include support for the necessary markup in texvc ;-). etc.

> Image layout and table support is currently available on Wikipedia with
> HTML syntax

Yes, but this is suboptimal because only a relatively small subset of
Wikipedia users knows HTML, and for those who don't, the HTML syntax is
unnecessarily complex. (We also get lots of different table styles because
some HTML wizards love to experiment.) That's why we use the wiki syntax
instead of HTMl all over the place. But tables are tricky, therefore we
haven't implemented them yet (some good proposals exist).

Regards,

Erik

I wrote in part:

>I intended to write an essay on texvc and how it *should* work
>on MetaWikipedia -- where such an essay belongs --
>once I got back online (today or later in the weekend),
>although I didn't know (until reading my email just now)
>that I would have this impending deadline to delay its implementation.
>I'll try to get it done today (not just sometime this weekend).

Well, Wikipedia was down when I wanted to do some testing,
and then I was pretty busy Saturday --
but I stayed up all night to write it!
So see [[m:texvc]] (where much other discussion of texvc could go;
there is already a place for Taw to explain it himself).

http://meta.wikipedia.org/wiki/texvc


-- Toby