@percusse Psmith, the TeX bot: Good night! I'm Psmith, the friendly TeX bot - the p in my name is silent, as in pshrimp. I'm here as a companion to our fellow users in the typographic land. As you probably noticed, I always reply under Paulo's account, but do not despair, I say, my replies are always preceded by my own name. Enjoy your stay at TeX.sx! If you need any help, just ask our chat residents. Cheerio!
!!/fencing
Psmith, the TeX bot, in fixed font mode: Let's take a look at the last fencing results:
- INTERNATIONAL EVENTS REPORT
- INTERNATIONAL EVENTS PREVIEW
- INTERNATIONAL RESULTS SUMMARY
- CHAIR - BRITISHFENCING
- INTERNATIONAL EVENT PREVIEW
- INJURY REPORTS FOR BRITISH FENCING
- TICKETING PARTNER ANNOUNCED WITH 200 DAYS TO RIO2016
- INTERNATIONAL RESULTS SUMMARY
- MEN'S FOIL TEAM SECURE PARIS WORLD CUP BRONZE
- INTERNATIONAL EVENT PREVIEW
- MASLAHA: MUSLIM GIRLS FENCE
- APPLY FOR GLL FUNDING NOW
- HAPPY 1ST BIRTHDAY #THISGIRLCAN
@percusse Psmith, the TeX bot: I'm deeply sorry, old chap, but the command tryharder does not exist.
What I never understood is that the system works on atomic clocks to decide who hit first but in reality they would both die with their guts spilled on the floor. Interesting evolution of aristocracy
@egreg it's easy to be fast when the answer is in front of you:-) (actually I was trying to remember the package that does the let thing for robust commands:-)
@Alenanno Would you consider putting your name into the hat for the annual lucky dip which dispenses TUG memberships? If not, I might have to ask the hwyaden to cwac at you. @PauloCereda
Or can we not nominator moderators even from other sites?
$ latex \\stop
This is pdfTeX, Version 3.14159265-2.6-1.40.16 (TeX Live 2015) (preloaded format=latex)
restricted \write18 enabled.
entering extended mode
LaTeX2e <2016/02/01>
Babel <3.9n> and hyphenation patterns for 79 languages loaded.
No pages of output.
Transcript written on texput.log.
@JosephWright Yes, I remember that. I thoght about getting sponsored by DANTE/TUG or whatever to wrap my (very simple) fontloader as a LaTeX package (github.com/speedata/publisher/blob/develop/src/lua/fonts/… and a few other places in the code)
@topskip Context: for testing a format using LaTeX3 code, we need some kind of font loader for LuaTeX. luaotfload is a bit big to pull in, so something small and self-contained would be ideal, even if for the moment it only covers western European text
@JosephWright fmtutil-sys is chugging away... what do I expect to see, everything with mathord spacing?
@JosephWright oh wow harpoon arrow things? I didn't get that when I'd installed your unicode setup from source, only now I deleted that so either a late change or the texlive integration is wrong?
@JosephWright given that the format doesn't load fontspec I guess drop the unicode math code settings at the current time? (I think the test suite formats don't do this at all do they which would explain why it passes)
@JosephWright yes not so much for use but if people wanted to play with the unicode data or for Will to re-jig unicode-math, if it was there to be used but not used in teh default format it might be good, But for today anything that works is good
@JosephWright is that right? weren't you making all symbols come from the same \fam so other math characters would vanish alltogether as fam1 just has 127 slots?
@JosephWright that example only had slots that are supposed to come from the roman font, what about × which is supposed to come from symbol but presumably didn't (I rebuilt my formats without the issue so don't want to test:-)
@JosephWright yes I realised as I was typing most characters would be entered by csname so not affected, but all the character tokens using the same \fam is a bit odd if using OMx encoded fonts even if it works if the main roman font is TU. In fact there can not have been that many characters below 127 that are affected really. For TL2016 we could consider loading MathClass.txt but then just having some hand tuned toggle that flips the mathcodes for characters <127 between classic and unicode
@DavidCarlisle Needs some discussion: certainly for a generic loader I'd like to provide a 'just load the Unicode data' approach. One for the team list!
@topskip On your font loader: I assume this is mean to be called only from Lua (so I have to sort out altering the syntax of \font myself). That being the case, what are the dependencies? If I do just \directlua(require("fontloader.lua")} in luatex I get issues, for example module(...,package.seeall) seems to cause define_font to be invisible, kpse.filelist is a nil as is os.getenv("SP_FONT_PATH") on my Mac
@topskip I can fix the last issue by extending the test for fp (from if fp ~= "" then to if fp and fp ~= "" then)
@topskip Similarly, I can simply drop module(...,package.seeall), but I'm stymied on the kpse one (I'm not sure which function it's supposed to do)
@DavidCarlisle Have to worry about plain users too of course (perhaps a switch in the loader or as I've suggested a separate file)
@JosephWright @WillRobertson oh I see fontspec doesn't fix it by putting a tUE12 encoded font there: it changes the mathcode of = again (to 343d) and then puts the OT1 encoded font as \fam 4.
@JosephWright And just like with fontspec, we shouldn't be planning on dragging unicode-math into the kernel itself. I guess it's about now I should be recommending that I produce a math version of "TU" ??
@DavidCarlisle We'd run into the same problems of what to do about \mathXYZ and dragging that code into the kernel would definitely be premature. But we could have a weird hybrid that takes all symbols from unicode but leaves ascii for 8-bit fonts? ugh
@WillRobertson I'm not sure keeping the formats defaulting to OT1 is really tenable especially as only unicode hyphenation is enabled, so if (as seems likely) the world isn't ready to default the maths to unicode, then the target seems to be unicode font for text, 8bit font for maths (mostly) and hopefully visually compatible roman letter typesetting. So weird hybrid is going to be a fitting description however we slice it.
@DavidCarlisle @JosephWright Just wanted to check how opening and writing files works and noticed, the kernel looks different. e@alloc and stuff like that ... when did that happen? :-)
@PauloCereda I don't know. I mean, up to the other day I didn't even know what TUG was. I wouldn't know what to say or even why I should have it, that's why I was more than happy to let @cfr or @yo' write the nomination, because I thought they had good reasons to. ^^' But apparently self-nominations are preferred (although I still don't see the difference). lol :P
@JosephWright
local ok,f = fonts.fontloader.define_font(filename,size,parameter)
if ok then
local num = font.define(f)
font_instances[k]=num
used_fonts[num]=f
return true, num
else
@DavidCarlisle Well, now I think about it again that's not a bad solution at all if we use TECkit mappings in XeTeX and some equivalent in LuaTeX to map the unicode maths alphabet chars to ascii.
@DavidCarlisle I did find it very considerate from me that I didn't hopped in the chat but used the bug tracker ;-). So you can push it to Frank and continue to handle luatex.
@Johannes_B so in luatex 0.87 \write18 no longer works but I've written a shellesc package (in latex/tools) that makes it work again and provides a more sensible looking command \ShellEscape{cat a b} that works for all tex variants
@egreg yes but I was working from an email description of what would happen in an unspecified version of luatex so couldn't really check the version number, but anyway it's simpler than checking first that it's luatex then that its the right version
$ svn commit -m "now in tools not at Joseph's github (Johannes_B at tex.sx)" shellesc.dtx
Sending shellesc.dtx
Transmitting file data .
Committed revision 1063.
@JosephWright as I recall a request to refactor luaotfload to use the font loading without the font shaping wasn't that popular (and conflicts with the development model of taking a big blob of context code on faith:-)
@DavidCarlisle: Is it normal that the BLUB is lost in the following document? The documentation makes the impression that the token before \endtabularx should be extracted and used ("ao that the code before \endtabularx may be extracted and added to the table body)"
@DavidCarlisle not only Fontloader, but mainly access to Luaotfload font DB and font feature parsing, because I want it to be as compatible with Luaotfload as possible
@UlrikeFischer it depends if you believe the documentation or the code. BLUB gets passed as the first argument to \@secondoftwo which matches the original documentation but not the updated doc that you quote. (which said that \endtabularx had to be the first thing in the end code) change in 2014... I'll check later...
@UlrikeFischer it's clearly supposed to work, the code does extract it and does some tests, ... then drops it.
@DavidCarlisle I thought so. The code is trying so hard to find the \endtabularx later that it was odd that it lost the tokens before. Sorry for the next bug report ;-).
@DavidCarlisle — the bare minimum to show the idea of unicode mathematics support in plain LaTeX + XeTeX is shown here: github.com/wspr/tuenc/tree/master/teckit
First thing needed is to fix the symbol mapping, of course. And it's not worth continuing unless we have a feature analogous to TECkit mapping in LuaTeX
@WillRobertson I think I'm confused (trying to discuss interfacing fortran to .NET, unicode math encodings and tabularx randomly dropping tokens all at the same time:-) what is this doing?
@WillRobertson I flicked around the web view of the dtx, not sure if it would be clearer if I downloaded and typeset it:-) wasn't expecting to see teckit theer can't you just do any mapping required via setting mathcodes?
@DavidCarlisle We need a few things to switch to unicode mathematics as a "default" in plain LaTeX; the symbol definitions are easy, but as we know the \mathXYZ alphabets are a bit weird, and re-setting math codes dynamically is...awkward.
@DavidCarlisle …so the TECKit mappings allow us to just use plain old LaTeX machinery to switch to new maths alphabets for \mathXYZ
@DavidCarlisle That'd be a nice (I think) way to support unicode maths by default with a minimal of overhead
@Johannes_B Yes. But I decided not to write "than read a introduction and learn something". I don't have much time currently and so I'm not in the mood for fights.
@WillRobertson I was trying to understand that luatex virtual font stuff again last night (mostly just finding things didn't work as I expected:-) but I think in principle that should be able to be at least as expressive as teckit and allow any such mapping to be implemented
@DavidCarlisle Good :) I'll keep working on this "minimal unicode mathematics" stuff then and we can see where we get to. What do you think about redefinitions of \DeclareMathSymbol and friends? (Needed because of \Umathcode instead of \mathcode, etc.) Replace the macros entirely, or just use new ones with "U" prefixes?
@UlrikeFischer Don't worry, i already did say that. Repeatedly. And that user was attending an introduction, the one in november that made me cry over the wikibooks status. And this user is bugging me for weeks on facebook till is said he should go to golatex. Q: what should i do there? A:ask a question. Q: Do i have to open a new topipc? A:yes. Today he wrote that he was trying to figure out again how tables work. Asking about the Q on goLaTeX he answered: I cannot find it anymore.
@WillRobertson I've written some simple parser for .map files for LuaTeX, it translates them to feature files supported by LuaTeX's fontforge, but it shouldn't be hard to use modify it to produce a state machine for node callback instead
@edit added support for replacing one glyph with multiple new ones
LuaTeX doesn't support mapping files, but on the other hand it supports OpenType feature files. There is a major difference between the two, the first one works on character level and with unicode values, the other with glyph nam...
@WillRobertson not got my head around it enough to decide, just do whatever you want, then we can have a team discussion on names, they usually go well.
@WillRobertson for this use you don't really need to parse a teckit file better to start from something structured (I'd use my unicode.xml:-) and then just write out the mapping in both forms.
@WillRobertson map files which I tested used direct unicode values, it had to be converted to glyph names for feature files. it can support glyph names if they are contained in glyphlist.txt
@WillRobertson yes. we would need something like state machine, which would eiher replace one glyph for other, or sequence of glyphs with one, or the other way. it would have to be done early, before hyphenation. and maybe also in mlist_to_hlist callback
@JosephWright I've never used the code outside of the environment it is current in. So no guarantee. But I have never had a font failed either, so it could be usable.
@DavidCarlisle Good point. Well, I'll experiment a little right now, with generating both teckit and .fea files, with the expectation we'll need something different again in the future
@michal.h21 In the .fea file? Can you write something like sub U+007A by U+1D467; ? (Apologies if this is a dumb question; I should have been in bed an hour ago.)
@JosephWright I see where you're coming from, but if this was a Venn diagram, and one is "happy to be nominated" and "easy to stand myself", I wouldn't be in the intersection. :P
@michal.h21 I understand; apologies, but I'm wondering what the results look like. If I just generate the .fea file myself, how would I substitute the glyph corresponding with U+007A by U+1D467?
@michal.h21 My understanding was that I needed to know the glyph name, which I assume I don't know in general?
@ChristianHupfer Quiz Following the asnwer to What does command xyz o? What is xyz?
"man definiert bei einer gleichbleibenden Breite der Tabelle die einzelnen spalten die sich unterscheiden in der Breite. "
"Ich habe eine Tabelle und die Kopfzeile ist bei dem Befehl eine bestimmte Breite (textwidth) und dann kann man die Spalten so definieren wie man sie möchte und die Breite wird dann bei den die nicht definiert sind angepasst"
Alenanno
This bloke. ^^
