the reasond ($$) is clear, but that's quite inconsistent.
@daleif In `mathtools` there are some redundancies for \AtBeginDocument{ \providecommand*\dblcolon{\mathrel{\vcentcolon\mkern-.9mu\vcentcolon}} \providecommand*\coloneqq{\mathrel{\vcentcolon\mkern-1.2mu=}} \providecommand*\Coloneqq{\mathrel{\dblcolon\mkern-1.2mu=}} \providecommand*\coloneq{\mathrel{\vcentcolon\mkern-1.2mu\mathrel{-}}} \providecommand*\Coloneq{\mathrel{\dblcolon\mkern-1.2mu\mathrel{-}}} \providecommand*\eqqcolon{\mathrel{=\mkern-1.2mu\vcentcolon}} \providecommand*\Eqqcolon{\mathrel{=\mkern-1.2mu\dblcolon}}
@daleif as \dblcolon, \vcentcolon, and clearly \mathrel{-} are all math relatives, the first \mathrel is not needed. Thanks.
but frankly, TeX would use some better logging features. Like a switch to clean up the log from things people do not understand anyway. TeX-specific editors do help, probably, to some extent
@LaRiFaRi yes probably misread the thread a bit: it was just ending as I came back to my desk:-) Are you planning to mail the luatex list (or use their bug tracker) or I could. It seems to affect new commands like \Umathchardef as well as classic \mathchardef defined tokens
The parameters for the mathematical model can be obtained by laboratory experiments, operating values analysis from the information system or measurement on the device. \\ \textbf{Determination of heat transfer coefficient for the mathematical model} \\ Magnesite samples ... 300 $\degree C$ (see Fig.~\ref{fig:fig6}). The enter of .... With the temperature of ~ 20 $\degree C$, the samples were gradually ... individual samples was <0.5 $mm$ with a volume of about 30 $cm^{3}$.
"The commands \mathchar , and \Umathchar and control sequences that are the result of \math- chardef or \Umathchardef are also acceptable in the horizontal and vertical modes. In those cases, the \textfont from the requested math family is used."
@LaRiFaRi the main loss is that it loses the error check that traditionally you get if you forget $, but they would probably say making zzz \alpha zzzwork rather than give an error is a plus. Personally I'd prefer the error but on error free documents there is no difference so hard to make a conclusive example where it does harm.
@DavidCarlisle Well, you could reply to that mailing you have found in order to get an explanation for that 'feature'. Maybe they have thought of something usefull.
@LaRiFaRi I could, put perhaps I won't, we are gradually trying to get latex to a position where we can support luatex as an engine and that means we need at times to report real bugs that are stopping that happening, so try to keep my bug reports to a minimum to highlight those cases:-)
@DavidCarlisle A clear loss in my eyes is the compatibility to the other formats. Thinking about that, I think it does a real harm. It should be no problem to use \textalpha or some \ensuremathes where it is really needed and desired
@LaRiFaRi one thing we found very quickly is that "compatibility with other formats" is a non-issue. For latex it is a major issue, but it is not part of the design aim for luatex/context
@DavidCarlisle In my example above \widthof{\equiv}, it could have done harm because the spacing of the text version could be different than what I expect. With the error, I would have found my clear mistake at once.
@DavidCarlisle What a pity. I am using Lua... all the time but would like to copy parts of it to somewhere else, still. Well, I will have to keep my coding clean than. Thanks for looking at it and, yes, I understand your remarks on bug-reporting.
@LaRiFaRi yes (but playing devil's advocate) the difference with classic tex is that in classic text your text font has no such symbol so an error is best that could be done anyway, but with unicode fonts your text font probably has got U+2261 so there is an argument that making \equiv access that character is a reasonable thing to do.
ah, ok. I ensuremath or text quite a few things. Handy sometimes if you know what you are doing. I always try to keep the things looking the same. That's why I stated above that I would not have liked to get \equiv from some other font.
@JosephWright OK need to check something else first I seem to have broken something \end occurred when \ifnum on line 150 was incomplete) had that for a while when testing but thought I'd fixed it before commit.... (unpack.ins)
@LaRiFaRi No, it will promote wrong usage: no error means that \alpha+\beta becomes legal in text mode and I've seen so bad input on Math.SE that this worries me.
@JosephWright I... have no opinion at this stage :) I can see the arguments both ways (well, playing Devil's Advocate for Karl) and neither is enough — yet — to convincingly sway me either way. I definitely agree that dropping the prefixes is cleaner. And LuaLaTeX is so young, and the chance of name clashes small enough, that I think the jump is a good idea.
@egreg I've thought about this before in the context of "what happens when users start pasting in plain unicode math" — you might think it's nice to auto-detect when maths starts and switch modes without needing markup. But the same argument can be made for auto-language switching when the script changes — it's too "leaky" around the edges. Sure, "where ratio \alpha is bleah" works fine, and maybe even "sum of ratios \alpha+\beta is blah" works too…
@egreg … but what about "where -\gamma causes instability" ? You can't parse it reliably so you should always require the math switch in the first place. (You might make a very similar argument about breqn!! But I think the situation there is slightly different. For reasons that escape me at present.)
@egreg Oh, what I mean is that once \alpha is seen as a math symbol, the mode remains in maths mode until "text" reappears (the same technique used by \XeTeXcharclass). It definitely doesn't work now :)
@egreg, @WillRobertson as noted above \alpha+\beta becomes legal in text mode with unicode-math even in xetex so not worrying too much about luatex primitive, making it an error in uniocde-math would be quite expensive I suspect.
@DavidCarlisle I had never thought through the ramifications of that. Hmm.
@DavidCarlisle I could make all symbols math active (which is definitely under consideration for the longer term) and set them to the invalid catcode otherwise?
@egreg I have to read entire books in which not a single italic (let alone math italic) is seen for text that's full of mathematics. Makes me sad.
@DavidCarlisle Ohhhhh, well users should be typing in plane 1 unicode. And I could adjust the macros that expand to greek plane 0 to expand to a mathchar instead? But then still have the LuaTeX issue :)
@DavidCarlisle s/mathchar/mathchardef'd symbol/
@DavidCarlisle In the end it comes down to the same problem as ASCII :) If the users type the wrong thing, we can't always save them.
@WillRobertson I suspect so. repeating the \alpha,\textalpha duplication over the entire unicode range probably didn't appeal:-)
@WillRobertson but while making greek letter commands work like textual greek is reasonble, it's a shame that it affects everything such as \sum and, as noted above, \equiv. But most real math expressions hopefuuly have ^ or \left or something that will generate an errror in text mode
@DavidCarlisle, @WillRobertson I'll put in a few pull requests to start tackling the \luatex... business in a 'safe' way (leave open a change without breaking anything now)
@WillRobertson, (@DavidCarlisle) Rather than adjust fontspec, unicode-math, etc. to look for the primitive names, a better plan is to require an up-to-date expl3 as I've already set that up to collect them up 'safely' as \luatex_... or \utex.... I'll make some pull requests using that approach later on (after work).
@WillRobertson Probably ignore the latest pull request until I sort this :-) (or I'll sort myself anyway in a later one)
The dawn of a new year, 2015, now approaches, or has already approached, either way it means that it is now time to reset our Community Promotion Ads!
What are Community Promotion Ads?
Community Promotion Ads are community-vetted advertisements that will show up on the main site, in the right s...
I'm just blind right now, this might be basic, sorry if it's a silly question. Why does =\mathrel{\mathop:}= vertically center the colon while =\mathrel{\mathop{:}}= doesn't?
@Manuel actually I misread your example a bit. \mathop is defined to vertically centre its content if it is a single character so {} or anything really, stops the centering. Let me just check exact;y what tex's doing on that example (I thought the first was an error to be honest:-)
@DavidCarlisle It seems weird to me, but I don't know what's happening.
In any case, mathtools has this in account. The definition of \vcentcolon is \mathrel{\mathop\ordinarycolon} which in turn does \mathrel{\mathop:}. But seems weird anyways.
Another question, for myself, is why did I think that braces were mandatory :)
@Manuel texbook says `^|\mathord|, ^|\mathop|, ^|\mathbin|, ^|\mathrel|, ^|\mathopen|, ^|\mathclose|, and ^|\mathpunct| are used for this purpose; each of them is followed either by a single character or by a subformula in braces.`
@FrankMittelbach a bit arbitrary I think since the start is sort of unnamed. begin/end and start/stop go together but since the start is just exp:w that doesn't really give much of a hint to the name of the end command.
@Johannes_B hard to say! perhaps, removing a "bad" sentence by editing, and PM the poster that it's off topic and may cause trouble, in a friendly positive way
@JosephWright ah RequireLuaModule could go then it probably wouldn't need luatexbase and would work with ltluatex (although that's academic until luaotfload doesn't force luatexbase)
although scratch that for now the lua code really depends on the existing luatexbase lua code, oops `require("luatexbase.modutils") require("luatexbase.cctb")`
@JosephWright I suppose we need to search for other uses of require("luatexbase.zzzz") in texlive and see about updating them....
@JosephWright I think so looking by eye, haven't downloaded your version to really test. longer term I suppose we'd want to remove the luatexbase emulation dependency but that can wait?
@JosephWright It's probably possible to write the code so it works with either but without any easy way for anyone other than us to test that probably not reasonable to make a pull request that does that,
@JosephWright well that to start with but I notice it uses the error functions returned from the module declaration so that would need adjusting ...
@JosephWright for similar reasons I suspect that luaotfload won't change until after a release so they can test the code on an existing format.
@JosephWright oh actually lualatex-math doesn't use them it just declares them local err, warn, info, log = luatexbase.provides_module({ so I guess that's OK, they are all just nil.
@JosephWright just thinking..... could do some fancy lua or chain that picked up the right thing, although tempting to do the test in tex, ...
@JosephWright untested but possibly change local cctb = luatexbase.catcodetables to local cctb = luatexbase.catcodetables or {string=registernumber('catcodetable@string')} should work?
Expansion question: How do I get percent signs but not backslashes in the following? \ClassWarning{syllabus}{Total assignment values (\the@TotalValues\%) must equal 100\%}