setuptex changes in minimals (was: bugs in beta)

newer
lots of luatex process/instances...

older
\setbuffer

Mojca Miklavec

8 May 2009 8 May '09

5:19 a.m.

On Fri, May 8, 2009 at 11:46, Yanrui Li wrote:

...

Recently I have found some environment variables in setuptex script are commented out sometimes, such as TEXMFLOCAL, TEXMFCONTEXT and so on. Is this be done deliberately?

Yes (we were bored during some Polish talks at BachoTeX). The reason for setuptex setting all those variables in the old days was to override tetex's environmental variables, but in principle this is not needed at all unless tetex is present on computer. The reason for removing them again was to make sure that Hans' scripts work properly even when no environmental variable is set (which will be the case on TeX Live for example). The only variable change that's really needed is fixing PATH, but it's so convenient to be able to run ". /path/to/setuptex" instead of typing the long export PATH="/path/to/minimals/bin:$PATH" We nevertheless left two or three variables since ConTeXt didn't yet work properly without them, but we'll remove those as well. If anyone uses minimals on tetex, we can create an additional file again that would enable complete isolation. We didn't clean up the windows script yet as it would need some more testing (and we had enough problem by minimals not running at all on unix), but that's still on TODO list ... We also silently added TEXMFHOME (while Hans was not watching), but maybe we need to change the exact path. If you have any particular question or request about that, let us know. Mojca

Show replies by date

Yanrui Li

8 May 8 May

6:16 a.m.

On Fri, May 8, 2009 at 7:19 PM, Mojca Miklavec wrote:

...

On Fri, May 8, 2009 at 11:46, Yanrui Li wrote:

...
Recently I have found some environment variables in setuptex script are commented out sometimes, such as TEXMFLOCAL, TEXMFCONTEXT and so on. Is this be done deliberately?

Yes (we were bored during some Polish talks at BachoTeX).

The reason for setuptex setting all those variables in the old days was to override tetex's environmental variables, but in principle this is not needed at all unless tetex is present on computer. The reason for removing them again was to make sure that Hans' scripts work properly even when no environmental variable is set (which will be the case on TeX Live for example).

The only variable change that's really needed is fixing PATH, but it's so convenient to be able to run ". /path/to/setuptex" instead of typing the long export PATH="/path/to/minimals/bin:$PATH"

We nevertheless left two or three variables since ConTeXt didn't yet work properly without them, but we'll remove those as well.

If anyone uses minimals on tetex, we can create an additional file again that would enable complete isolation.

We didn't clean up the windows script yet as it would need some more testing (and we had enough problem by minimals not running at all on unix), but that's still on TODO list ...

We also silently added TEXMFHOME (while Hans was not watching), but maybe we need to change the exact path.

If you have any particular question or request about that, let us know.

Thanks for your explanation. Now I see. I have no questions for these, only I have to modify instruction of these variables in the tutorial which I wrote for Chinese users :) -- Best wishes, Li Yanrui

Mojca Miklavec

11:43 a.m.

On Fri, May 8, 2009 at 14:16, Yanrui Li wrote:

...

I have no questions for these, only I have to modify instruction of these variables in the tutorial which I wrote for Chinese users :)

What exactly do you describe there? Do you write instructions for users that want to modify their installation? Mojca

Yanrui Li

12:17 p.m.

On Sat, May 9, 2009 at 1:43 AM, Mojca Miklavec wrote:

...

On Fri, May 8, 2009 at 14:16, Yanrui Li wrote:

...
I have no questions for these, only I have to modify instruction of these variables in the tutorial which I wrote for Chinese users :)

What exactly do you describe there? Do you write instructions for users that want to modify their installation?

MkIV have not support Chinese typesetting now, so I modified scrp-ini.lua and scrp-cjk.lua for compression of Chinese punctuations. In addition, I wrote a lua program which can solve the problem of margin alignment of Chinese puncuations appeared on the right side. To simplify the usage of these modified script files, I used TEXMFLOCAL instead of the specific path. But this is not a very serious matter, because I am going to rewrite the related content easily and only a few Chinese users are interested in MkIV . -- Best wishes, Li Yanrui

Yanrui Li

12:20 p.m.

On Sat, May 9, 2009 at 2:17 AM, Yanrui Li wrote:

...

On Sat, May 9, 2009 at 1:43 AM, Mojca Miklavec wrote:

...
On Fri, May 8, 2009 at 14:16, Yanrui Li wrote:

...
I have no questions for these, only I have to modify instruction of these variables in the tutorial which I wrote for Chinese users :)

What exactly do you describe there? Do you write instructions for users that want to modify their installation?

MkIV have not support Chinese typesetting now, so I modified scrp-ini.lua and scrp-cjk.lua for compression of Chinese punctuations. In addition, I wrote a lua program which can solve the problem of margin alignment of Chinese puncuations appeared on the right side. To simplify the usage of these modified script files, I used TEXMFLOCAL instead of the specific path. But this is not a very serious matter, because I am going to rewrite the related content easily and only a few Chinese users are interested in MkIV .

Sorry! s/MkIV have not support Chinese typesetting now / MkIV have not support Chinese typesetting well now/g -- Best wishes, Li Yanrui

Yue Wang

7:57 p.m.

On Sat, May 9, 2009 at 2:20 AM, Yanrui Li wrote:

...

Sorry!

s/MkIV have not support Chinese typesetting now / MkIV have not support Chinese typesetting well now/g

We can start a fork of scrp-* on code.google.com/p/ctex-kit, making a usable version, and later ask Hans to merge back into ConTeXt. Yue Wang

...

-- Best wishes, Li Yanrui ___________________________________________________________________________________ If your question is of interest to others as well, please add an entry to the Wiki!

maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context webpage : http://www.pragma-ade.nl / http://tex.aanhet.net archive : https://foundry.supelec.fr/projects/contextrev/ wiki : http://contextgarden.net ___________________________________________________________________________________

Hans Hagen

9 May 9 May

2:02 a.m.

New subject: setuptex changes in minimals

Yue Wang wrote:

...

On Sat, May 9, 2009 at 2:20 AM, Yanrui Li wrote:

...
Sorry!

s/MkIV have not support Chinese typesetting now / MkIV have not support Chinese typesetting well now/g

We can start a fork of scrp-* on code.google.com/p/ctex-kit, making a usable version, and later ask Hans to merge back into ConTeXt.

it all depends how scrp evolves as it has to deal with more issues too; i will not merge code that i don't fully understand but experimenting does not hurt; one problem with script stuff is that (in the future) it might be closely cooperate with other mechanisms of context (it used to be part of the font analyser, now it's plugged in elsewhere but even that might change, for instance when in upcoming versions of luatex we have a bit more control) also, some issues will be dealt with in other code in the future (i simply cannot touch all code at the same time) so don't expect mkiv to be finished before 2012 (when luatex 1 is ready); it's a stepwise process and it's only driven by context (i.e. generic code only shows up when context code works okay); anyhow, as long as nobody tells me what the expected behaviour is, with test code etc i can;t fix anything; one complication (e.g. with chinese) is that over the last few year is got conflicting requests; now, if there are multiple ways to deal with it, then i can implement variants but only if i know the exact conventions (with proper names for them) for instance when i found out that cjk-korean is not the same as cjk-chinese (one of them) i split the machinery so we're not talking cjk here, but very specific needs; after the split i never tested chinese (no samples) and no one else propbably did that either keep in mind that context is not meant for one language but for many and that when possible mechanisms need to be able to used mixed; there will never be exclusive solutions for one languages unless it's a very specialized module anyhow, we can best follow the same route as with other language / script support - we need consistent specs, or maybe multiple specs as variants - as well as test files for each such case - and then we can look into what can be improved or added Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------

Yanrui Li

5:44 a.m.

New subject: setuptex changes in minimals

On Sat, May 9, 2009 at 4:02 PM, Hans Hagen wrote:

...

Yue Wang wrote:

...
On Sat, May 9, 2009 at 2:20 AM, Yanrui Li wrote:

...
Sorry!

s/MkIV have not support Chinese typesetting now / MkIV have not support Chinese typesetting well now/g

We can start a fork of scrp-* on code.google.com/p/ctex-kit, making a usable version, and later ask Hans to merge back into ConTeXt.

it all depends how scrp evolves as it has to deal with more issues too; i will not merge code that i don't fully understand but experimenting does not hurt; one problem with script stuff is that (in the future) it might be closely cooperate with other mechanisms of context (it used to be part of the font analyser, now it's plugged in elsewhere but even that might change, for instance when in upcoming versions of luatex we have a bit more control)

What I have done is only experiment. I will try my best to follow the evolves of the scrp-* files until that day, when MkIV support Chinese typesetting well, comes. I have to introduce MkIV to Chinese users and make them paying close attention to it in this way.

...

also, some issues will be dealt with in other code in the future (i simply cannot touch all code at the same time) so don't expect mkiv to be finished before 2012 (when luatex 1 is ready); it's a stepwise process and it's only driven by context (i.e. generic code only shows up when context code works okay);

anyhow, as long as nobody tells me what the expected behaviour is, with test code etc i can;t fix anything; one complication (e.g. with chinese) is that over the last few year is got conflicting requests; now, if there are multiple ways to deal with it, then i can implement variants but only if i know the exact conventions (with proper names for them)

Yes. It is a key reson that we are still not familiar with MkIV well now. But with in-depth understanding of MkIV, maybe many suggestions will be post to you. Of course these suggestions should be collected according actual demands by Chinese TeX users group.

...

for instance when i found out that cjk-korean is not the same as cjk-chinese (one of them) i split the machinery so we're not talking cjk here, but very specific needs; after the split i never tested chinese (no samples) and no one else propbably did that either

keep in mind that context is not meant for one language but for many and that when possible mechanisms need to be able to used mixed; there will never be exclusive solutions for one languages unless it's a very specialized module

anyhow, we can best follow the same route as with other language / script support

- we need consistent specs, or maybe multiple specs as variants - as well as test files for each such case - and then we can look into what can be improved or added

I think that maybe this needs a very long time until CJK users know what each other needs. But three years should be enough :) -- Best wishes, Li Yanrui

Yue Wang

9:57 a.m.

New subject: setuptex changes in minimals

Hi, list: Well, TeX is evolving rapidly these years (especially in the LuaTeX world). So there are many things that Chinese typographers never imagine about in the past can now be solved thanks to LuaTeX's development. For example: - Now we can use Lua to detect the character code. so it is possible to assign different fonts for latin, CJK, CJK-ext A, CJK-extB. font switching is easy and elegent. This is not possible in tex82. - It is ugly to produce a paragraph whose last line only consists one Chinese character. In order to avoid that, tex82 programmers should write very complicated macros. So many developers here tend to ignore this problem. However, in LuaTeX, this is quite easy. It is simple to scan the position before the punctuation and \par in lua, and insert a line break penalty there. - Punctuation compression. In the past this is not easy to do (a lot of macros should be defined), and because each Chinese fonts are so distinctly designed, their punctuations' positions in the character box are different. So different parameters should be applied when different fonts are involved in typesetting. However, LuaTeX make it possible to calculate character bounding box information, so punctuation compression is as easy as re-setting the punctuation width. There are many more examples, and these are just three. But we can see that with the development of TeX engine (especially the open-up of TeX engine), users might consider better ways for old solutions, and even request for possibilities they never thought about before. So it is possible for different users to provide different specifications. But we (me, Li Yanrui, Chen Zhichu) as the board members of the Chinese TeX Society, will try our best to organized the typesetting discussion on local forums, taking different opinions into consideration, and later provide a complete specification as long as we can. Chinese support in ConTeXt is not easy, we should thank all the developers who contribute to our language (esp, Hans, Taco and Wolfgang), their innovation and hardworking make Chinese typesetting in ConTeXt possible, and drastically improve the performance of font loading. Though the current solution is not perfect (eg. there are many more things we should fix and implement, and the current font loading time is still a bit too long), I think with the current fast pace, we will have a very prospective solution near luatex 1.0:) Yue Wang On 5/9/09, Yanrui Li wrote:

...

On Sat, May 9, 2009 at 4:02 PM, Hans Hagen wrote:

...
Yue Wang wrote:

...
On Sat, May 9, 2009 at 2:20 AM, Yanrui Li wrote:

...
Sorry!

s/MkIV have not support Chinese typesetting now / MkIV have not support Chinese typesetting well now/g

We can start a fork of scrp-* on code.google.com/p/ctex-kit, making a usable version, and later ask Hans to merge back into ConTeXt.

it all depends how scrp evolves as it has to deal with more issues too; i will not merge code that i don't fully understand but experimenting does not hurt; one problem with script stuff is that (in the future) it might be closely cooperate with other mechanisms of context (it used to be part of the font analyser, now it's plugged in elsewhere but even that might change, for instance when in upcoming versions of luatex we have a bit more control)

What I have done is only experiment. I will try my best to follow the evolves of the scrp-* files until that day, when MkIV support Chinese typesetting well, comes. I have to introduce MkIV to Chinese users and make them paying close attention to it in this way.

...
also, some issues will be dealt with in other code in the future (i simply cannot touch all code at the same time) so don't expect mkiv to be finished before 2012 (when luatex 1 is ready); it's a stepwise process and it's only driven by context (i.e. generic code only shows up when context code works okay);

anyhow, as long as nobody tells me what the expected behaviour is, with test code etc i can;t fix anything; one complication (e.g. with chinese) is that over the last few year is got conflicting requests; now, if there are multiple ways to deal with it, then i can implement variants but only if i know the exact conventions (with proper names for them)

Yes. It is a key reson that we are still not familiar with MkIV well now. But with in-depth understanding of MkIV, maybe many suggestions will be post to you. Of course these suggestions should be collected according actual demands by Chinese TeX users group.

...
for instance when i found out that cjk-korean is not the same as cjk-chinese (one of them) i split the machinery so we're not talking cjk here, but very specific needs; after the split i never tested chinese (no samples) and no one else propbably did that either

keep in mind that context is not meant for one language but for many and that when possible mechanisms need to be able to used mixed; there will never be exclusive solutions for one languages unless it's a very specialized module

anyhow, we can best follow the same route as with other language / script support

- we need consistent specs, or maybe multiple specs as variants - as well as test files for each such case - and then we can look into what can be improved or added

I think that maybe this needs a very long time until CJK users know what each other needs. But three years should be enough :)

-- Best wishes, Li Yanrui ___________________________________________________________________________________ If your question is of interest to others as well, please add an entry to the Wiki!

maillist : ntg-context@ntg.nl / http://www.ntg.nl/mailman/listinfo/ntg-context webpage : http://www.pragma-ade.nl / http://tex.aanhet.net archive : https://foundry.supelec.fr/projects/contextrev/ wiki : http://contextgarden.net ___________________________________________________________________________________

Hans Hagen

10 May 10 May

9:35 a.m.

New subject: setuptex changes in minimals

Yue Wang wrote:

...

- It is ugly to produce a paragraph whose last line only consists one Chinese character. In order to avoid that, tex82 programmers should write very complicated macros. So many developers here tend to ignore this problem. However, in LuaTeX, this is quite easy. It is simple to scan the position before the punctuation and \par in lua, and insert a line break penalty there.

ok, but it might even be more clever, like minimum number of glyphs on last line or so (and words in case of other scripts)

...

- Punctuation compression. In the past this is not easy to do (a lot of macros should be defined), and because each Chinese fonts are so distinctly designed, their punctuations' positions in the character box are different. So different parameters should be applied when different fonts are involved in typesetting. However, LuaTeX make it possible to calculate character bounding box information, so punctuation compression is as easy as re-setting the punctuation width.

i had code for that but removes it as fonts are pretty inconsistent; i will look into that again when we have (1) a couple of reference fonts without bugs in the tex distribution and (2) wiki pages with info about the quality of fonts (can later become a database); otherwise we end up in messy support situations

...

But we (me, Li Yanrui, Chen Zhichu) as the board members of the Chinese TeX Society, will try our best to organized the typesetting discussion on local forums, taking different opinions into consideration, and later provide a complete specification as long as we can.

that would be very helpful

...

pace, we will have a very prospective solution near luatex 1.0:)

5557

Age (days ago)

5559

Last active (days ago)

List overview

Download

9 comments

4 participants

participants (4)

Hans Hagen
Mojca Miklavec
Yanrui Li
Yue Wang

setuptex changes in minimals (was: bugs in beta)

tags

participants (4)