| 6 Oct 2021 |
|  nicfab left the room. | 12:53:21 |
|  Phelype Oleinik set a profile picture. | 13:00:11 |
 Joseph Wright | hmenke: I guess I could do with a little insight into the build setup for the manual. What's the reasoning behind having all those engine-dependent files? There's only one manual, so normally one sets up to typeset with a single engine | 13:07:22 |
| Joseph Wright | hmenke: I get the need for making an HTML version, but not e.g. xelatex/dvipdfm/dvips on top of pdflatex | 13:07:48 |
 hmenke | In reply to @josephwright:matrix.org
hmenke: I get the need for making an HTML version, but not e.g. xelatex/dvipdfm/dvips on top of pdflatex In the future these redundant engines can be dropped for compiling the manual (in fact, the only relevant engine for producing the manual is luatex), but right now this is PGF's poor man's testsuite. | 13:08:53 |
| hmenke | If the manual typesets with all engines, we probably didn't break anything. | 13:09:27 |
| hmenke | Not great, but that's the best we have right now. | 13:09:38 |
| Joseph Wright | In reply to @hmenke:matrix.org
In the future these redundant engines can be dropped for compiling the manual (in fact, the only relevant engine for producing the manual is luatex), but right now this is PGF's poor man's testsuite. Ah, in that case I'll focus on LuaTeX and assume that the tests will be moved elsewhere: that's the key | 13:09:40 |
| Joseph Wright | In reply to @hmenke:matrix.org
If the manual typesets with all engines, we probably didn't break anything. What I used to do with siunitx more or less | 13:09:52 |
| hmenke | Sounds reasonable. | 13:09:57 |
| Joseph Wright | In reply to @hmenke:matrix.org
Sounds reasonable. Aim after all is a PR that can be used to start a set of tests, so it doesn't need to get merged any time soon - get the setup right, make some tests, then merge across | 13:10:36 |
| Joseph Wright | hmenke: Getting there: I have typesetting of the manual now, a few rough edges but the shape is coming togehter | 15:01:53 |
| Joseph Wright | hmenke: How would you feel about me 'flattening' the file structure for the doc tree? It would make it easier to handle, also would fit with other packages, what goes to CTAN, etc. I will sort things one way or another, this is just the lowest energy pathway | 15:03:55 |
| Joseph Wright | hmenke: Ah, I spot something else: I get file copy issues for the graph drawing part of pgf. Looking at it, there are lots of files with the same names, and they are Lua files but in the tex/generic/pgf tree. Any idea if they are really 'live'? | 15:28:51 |
| hmenke | In reply to @josephwright:matrix.org
hmenke: How would you feel about me 'flattening' the file structure for the doc tree? It would make it easier to handle, also would fit with other packages, what goes to CTAN, etc. I will sort things one way or another, this is just the lowest energy pathway No | 15:34:47 |
| hmenke | In reply to @josephwright:matrix.org
hmenke: Ah, I spot something else: I get file copy issues for the graph drawing part of pgf. Looking at it, there are lots of files with the same names, and they are Lua files but in the tex/generic/pgf tree. Any idea if they are really 'live'? Yes, they are live. | 15:34:59 |
| hmenke | More detail later, got to go. | 15:35:08 |
| Joseph Wright | In reply to @hmenke:matrix.org
No Cool | 15:35:13 |
| Joseph Wright | In reply to @hmenke:matrix.org
Yes, they are live. Er, OK, that's odd ... I'll have to ponder a bit | 15:35:29 |
| Joseph Wright | In reply to @hmenke:matrix.org
Yes, they are live. I'll work on that later, then | 15:36:08 |
| Joseph Wright | hmenke: Meanwhile, I got some basic testing running as a demo: https://github.com/josephwright/pgf/commit/e29a119051aea4539ca44584ac0971faa89f9593 | 16:11:53 |
| Joseph Wright | hmenke: That part is much easier as l3build can definitely handle it - just a question of how much complexity you need in the test rig | 16:12:20 |
| Joseph Wright | hmenke: I will get things working such that it can do a full release too - I may need some new l3build features for that, but it should not be too much of a stretch | 16:12:55 |
| hmenke | josephwright: Now I have time. | 17:21:18 |
| hmenke | What did you have in mind when you asked about flattening the doc tree? | 17:21:34 |
| hmenke | Regarding the Lua stuff: Till used Lua the way Lua should be used, i.e. using directories for nesting modules. | 17:22:52 |
| hmenke | I think kpse is the only file lookup that cannot have files with the same name. | 17:23:16 |
| Joseph Wright | In reply to @hmenke:matrix.org
What did you have in mind when you asked about flattening the doc tree? Just that at present you have (at least) three places for the source files: text-en, version-for-luatex and images, and it would be easier to have all the files in one directory (and not then have subdirectories inside them) | 17:23:37 |
| Joseph Wright | In reply to @hmenke:matrix.org
I think kpse is the only file lookup that cannot have files with the same name. Hmm, I'm not completely convinced ;) | 17:24:05 |
| Joseph Wright | In reply to @hmenke:matrix.org
I think kpse is the only file lookup that cannot have files with the same name. It can have files with the same name, it's just you can't tell which is which - I think the same way as kpse I guess as for my day job I'd have unique filenames so they always map to one specific page in my lab book | 17:24:53 |
