Hi, So, for those who hesitate to check out lmtx, here is some information. - In order to achieve long term stability context will use a lean and mean variant of luatex (although for now context will keep running on luatex too; i might drop support for luajittex). Eventually the source code will be part of the context distribution so that one can always compile a binary that matches an archived context version. - This version also permits us to experiment (even very extreme) without interupting the parent program luatex, which is now used by other macro packages too and therefore needs to be frozen in functionality. Some features might trickle back into luatex eventually (unless it breaks compatibility). - As a direct effect it means that context does a bit more: creating the pdf file (already most is done anyway in mkiv), dealing with font embedding, handling image inclusion, etc. For sure there can be issues with that but it has run on my machine already for quite a while. - It is hard to be conclusive about performance but for instance the luatex manual processes a bit faster with lmtx as does the test suite. A gain in performance on very demanding documents is also observed by Alan and Thomas. Normally one can expect a similar performance. However, the memory footprint could be a bit smaller. - The current somewhat hybrid codebase of context will stepwise be adapted so that at some point we can start experimenting with some pending ideas. (Which is why i sometimes answer "later this year".) Anyway, it might not look like it, but it is a conceptual change in how context evolves. Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl -----------------------------------------------------------------
Hi,
On 2 Apr 2019, at 11:05, Hans Hagen
wrote: - It is hard to be conclusive about performance but for instance the luatex manual processes a bit faster with lmtx as does the test suite. A gain in performance on very demanding documents is also observed by Alan and Thomas. Normally one can expect a similar performance. However, the memory footprint could be a bit smaller.
For sure the ConTeXt now loads and initialises a lot faster. With that trivial issue.tex test file by Arthur I get these timings (after a few ‘warmup’ runs). MkIV (This is LuaTeX, Version 1.09.3 (TeX Live 2019/dev)): mkiv lua stats > stored bytecode data: 444 modules (0.445 sec), 93 tables (0.021 sec), 537 chunks (0.466 sec) ... mkiv lua stats > runtime: 0.888 seconds, 1 processed pages, 1 shipped pages, 1.126 pages/second system | total runtime: 1.391 seconds real 0m1.646s user 0m1.476s sys 0m0.115s LMTX: mkiv lua stats > stored bytecode data: 456 modules (0.243 sec), 93 tables (0.016 sec), 549 chunks (0.259 sec) ... mkiv lua stats > runtime: 0.559 seconds, 1 processed pages, 1 shipped pages, 1.789 pages/second system | total runtime: 0.655 seconds of 0.725 seconds real 0m0.731s user 0m0.601s sys 0m0.084s Impressive! Taco
On 2/04/19 10:05 PM, Hans Hagen wrote:
Hi,
So, for those who hesitate to check out lmtx, here is some information.
- In order to achieve long term stability context will use a lean and mean variant of luatex (although for now context will keep running on luatex too; i might drop support for luajittex). Eventually the source code will be part of the context distribution so that one can always compile a binary that matches an archived context version.
- This version also permits us to experiment (even very extreme) without interupting the parent program luatex, which is now used by other macro packages too and therefore needs to be frozen in functionality. Some features might trickle back into luatex eventually (unless it breaks compatibility).
- As a direct effect it means that context does a bit more: creating the pdf file (already most is done anyway in mkiv), dealing with font embedding, handling image inclusion, etc. For sure there can be issues with that but it has run on my machine already for quite a while.
- It is hard to be conclusive about performance but for instance the luatex manual processes a bit faster with lmtx as does the test suite. A gain in performance on very demanding documents is also observed by Alan and Thomas. Normally one can expect a similar performance. However, the memory footprint could be a bit smaller.
- The current somewhat hybrid codebase of context will stepwise be adapted so that at some point we can start experimenting with some pending ideas. (Which is why i sometimes answer "later this year".)
Where can I find the C sources for the luametatex engine?
Anyway, it might not look like it, but it is a conceptual change in how context evolves.
Hans
----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl ----------------------------------------------------------------- ___________________________________________________________________________________ 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://context.aanhet.net archive : https://bitbucket.org/phg/context-mirror/commits/ wiki : http://contextgarden.net ___________________________________________________________________________________
On 4/2/2019 10:11 PM, Henri Menke wrote:
On 2/04/19 10:05 PM, Hans Hagen wrote:
Hi,
So, for those who hesitate to check out lmtx, here is some information.
- In order to achieve long term stability context will use a lean and mean variant of luatex (although for now context will keep running on luatex too; i might drop support for luajittex). Eventually the source code will be part of the context distribution so that one can always compile a binary that matches an archived context version.
- This version also permits us to experiment (even very extreme) without interupting the parent program luatex, which is now used by other macro packages too and therefore needs to be frozen in functionality. Some features might trickle back into luatex eventually (unless it breaks compatibility).
- As a direct effect it means that context does a bit more: creating the pdf file (already most is done anyway in mkiv), dealing with font embedding, handling image inclusion, etc. For sure there can be issues with that but it has run on my machine already for quite a while.
- It is hard to be conclusive about performance but for instance the luatex manual processes a bit faster with lmtx as does the test suite. A gain in performance on very demanding documents is also observed by Alan and Thomas. Normally one can expect a similar performance. However, the memory footprint could be a bit smaller.
- The current somewhat hybrid codebase of context will stepwise be adapted so that at some point we can start experimenting with some pending ideas. (Which is why i sometimes answer "later this year".)
Where can I find the C sources for the luametatex engine?
nowhere ... as i mentioned these will become part of the context distribution around the ctx meeting (when we also have the compile infrastructure ready) what we offer now is an experimental test release Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl -----------------------------------------------------------------
On 3/04/19 9:19 AM, Hans Hagen wrote:
On 4/2/2019 10:11 PM, Henri Menke wrote:
On 2/04/19 10:05 PM, Hans Hagen wrote:
Hi,
So, for those who hesitate to check out lmtx, here is some information.
- In order to achieve long term stability context will use a lean and mean variant of luatex (although for now context will keep running on luatex too; i might drop support for luajittex). Eventually the source code will be part of the context distribution so that one can always compile a binary that matches an archived context version.
- This version also permits us to experiment (even very extreme) without interupting the parent program luatex, which is now used by other macro packages too and therefore needs to be frozen in functionality. Some features might trickle back into luatex eventually (unless it breaks compatibility).
- As a direct effect it means that context does a bit more: creating the pdf file (already most is done anyway in mkiv), dealing with font embedding, handling image inclusion, etc. For sure there can be issues with that but it has run on my machine already for quite a while.
- It is hard to be conclusive about performance but for instance the luatex manual processes a bit faster with lmtx as does the test suite. A gain in performance on very demanding documents is also observed by Alan and Thomas. Normally one can expect a similar performance. However, the memory footprint could be a bit smaller.
- The current somewhat hybrid codebase of context will stepwise be adapted so that at some point we can start experimenting with some pending ideas. (Which is why i sometimes answer "later this year".)
Where can I find the C sources for the luametatex engine?
nowhere ... as i mentioned these will become part of the context distribution around the ctx meeting (when we also have the compile infrastructure ready)
Hm. But luatex is released under GPLv2 and I reckon luametatex is based on luatex, i.e. it is a derived work which means that you legally *have* to publish the source.
what we offer now is an experimental test release
Hans
----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl -----------------------------------------------------------------
Hm. But luatex is released under GPLv2 and I reckon luametatex is based on luatex, i.e. it is a derived work which means that you legally *have* to publish the source. i publish when i feel ready to do (i.e. when it meets my quality criteria) ... arguments like the one above make me actually hate open
On 4/2/2019 10:30 PM, Henri Menke wrote: source ... ... in case you wonder: we don't make (and never made) money on lua(meta)tex, in fact all this development has always costs us money, time, etc, etc, and i won't let whatever license spoil this game that until now always has been mostly fun and supporting user demand Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl -----------------------------------------------------------------
participants (3)
-
Hans Hagen
-
Henri Menke
-
Taco Hoekwater