@DavidCarlisle We saw a gap in MathML in chrome. Exactly one rule width, so that should easily be fixed. But I do not remembering seeing something in mathjax.
low level Q: how does one cite a given commit on GitHub? I forgot to mention the issue id in my commit, so would want to cite the commit when I close it.
@CroCo it could be different build of chrome, different operating system, different locally installed fonts (this not so likely with mathjax) .... the web is a lot more hostile environment than tex which is designed to produce bitwise identical output on different platforms. On the web platform stuff changes and differences happen.
@DavidCarlisle I find web applications to be more stable than desktop apps. I have limited knowledge of JavaScript, as I mainly code in C/C++ and Matlab, which makes problem diagnosis more challenging.
@CroCo yes but tex isn't your average desktop app stability level, you can usually take a document from 1985 and process it with a current system and get identical output to the output obtained with a 1985 system.
Is there any reason why one does not have \bool_clear:N (alias for \bool_set_false:N), it would make the bool type more consistent with the various other types.
@DavidCarlisle It is not really the syntax highlighting I was after, more string completion (I probably just need to tell it that _ and : are parts of words.
anybody know what might cause ["ERROR", "Technical problem encountered. Please contact the web master"] for a ctan upload with l3build? the data in the curlopts file looks OK (though I don't know what it should look like - it just looks sane-ish). --debug produced a great deal of output, most of which isn't obviously meaningful. ctan lists the error, but doesn't say what to do - only that it means the upload data couldn't be saved.
@cfr sounds more like an issue at their side, you could just try again. (despite having written most of that upload code I hardly ever dare use it and usually use the web form, so I can enter one-off notes, but it should work....)
@cfr well if it's a sever issue, it may need to wait until someone finds some free time to fix it, you could try uploading the same zip via the web form that would tell you if the error is in the https post or in its validation of the zip
@Skillmon my question of course extend to those as well. Right know you need to look up/memorize which type having a clear and know what the clear equivalent is for the types that does not have it
The ought to be a \type_clear:N for all types that reset the value to the same value a var has right after \type_new:N for a consistent interface.
@daleif Actually, IIRC, there used to be \int_clear:N, but the team felt this wasn't right, because integer variables always have a “nonempty” value and, likewise, other variables such as dimensions, skips, fps. Thus _zero was introduced for these types. One can “clear” token lists, sequences and so on, meaning that they're set to empty.
@DavidCarlisle the main reason i tried l3build upload was because I invariably screw up the web form in some way which requires one of the overworked ctan people to chase me for corrections.
[I forget what I did last time and/or ctan change what they want.]
@DavidCarlisle it accepted the same zip via the web. presumably that does whatever checks are automated in the same way? but the data for the upload config in l3build doesn't seem to quite map onto the data required for the web form. (but shouldn't I have gotten some more meaningful error if that was the problem?)
@cfr sure but ["ERROR", "Technical problem encountered. means there was some unexpected error on the server so something crashed so you can't really expect a meaningfull high level error
@cfr yes or the server could have run out of disk space or dropped a network connection or... it might not have been anything wrong on your side at all
@DavidCarlisle either I'm missing something basic or, contra l3build's doc, repository can't be multiple ... if I substitute a single value, it validates, even if other fields have multiple values e.g. topic is ok,
but I guess l3build would also ideally not accept multiple values i.e. not just a doc issue?
@cfr yes the code has ` ctan_field("repository", uploadconfig.repository, 255, "URL(s) of source repositories", false, true )` and that final true means it takes multiple values, let me check what ctan says....
hmm ctan.org/help/submit implies it's a single value, I orginally cross checked against ctan's perl uploader, I thought but it was a while ago, probably l3build is wrong...
@cfr if you try ctan-o-mat --init to get a template uploader for ctan-o-mat then it says
% This field contains the version management.
% The value is optional.
% The value is a URL.
% The value is restricted to 255 characters.
% Multiple values are allowed.
\repository{}
allowing multiple repositories. I wonder if that actually works..
@cfr what error message do you get with a list of repos?
