« first day (3361 days earlier)      last day (1573 days later) » 

Faheem Mitha
Faheem Mitha
12:57 AM
@DavidCarlisle Oh. I thought it was just clickable links.
 
 
4 hours later…
Faheem Mitha
Faheem Mitha
5:27 AM
Is Martin Scharrer ever in chat?
 
David Purton
David Purton
5:50 AM
Is anyone around who owns pre-press tools, e.g., Acrobat Pro or something like PitStop Pro? I'm having trouble with a tikz fading. My printer says it's crashing his RIP and I can confirm that it will crash Acroread 9 under Linux when trying to output PostScript. @HenriMenke maybe?
 
CarLaTeX
CarLaTeX
6:48 AM
@FaheemMitha Last time I found searching for him: chat.stackexchange.com/transcript/message/43612256#43612256
 
Faheem Mitha
Faheem Mitha
@CarLaTeX Oh, that's a long time.
Thank you for the link.
 
CarLaTeX
CarLaTeX
@FaheemMitha The fact he is not active in chat doesn't mean he is not here, his user page says Last seen Jan 9 at 11:28 tex.stackexchange.com/users/2975/martin-scharrer
 
Faheem Mitha
Faheem Mitha
@CarLaTeX Yes, I understand that visiting the site and visiting chat are not the same things. :-)
I was hoping to ask him for his thoughts about transitioning from BitBucket.
I guess I could send him email.
 
CarLaTeX
CarLaTeX
@FaheemMitha Yes, I think it's better
Happy rubber duck day!
 
Faheem Mitha
Faheem Mitha
7:03 AM
@CarLaTeX Ok.
@CarLaTeX What's that?
 
CarLaTeX
CarLaTeX
 
Faheem Mitha
Faheem Mitha
@CarLaTeX Oh, I see.
I just ran into this - I don't see why hyperref can't deal with this.
22
Q: hash-tag in \href causes error

YotamI'm creating a beamer presentation. I want to put a link to a source which has a hash-tag in it (#). When I try to compile this I get an error: ! Illegal parameter number in definition of \test. <to be read again> } l.91 \end{itemize} % ends low level ? x ...

beamer hyperref
Odd. The hyperref manual says the following:
> The text is made a hyperlink to the URL; this must be a full URL (relative to the base URL, if that is defined). The special characters # and ~ do not need to be escaped in any way (unless the command is used in the argument of another command).
This text appears directly below: 
\href[options]{URL}{text}
This appears to contradict my experience. Am I missing something?
 
Skillmon
Skillmon
7:42 AM
@JosephWright no, currently active commas and equals aren't handled. And I wrote it expandable because a) that's a unique selling point, and b) it wasn't that much of an overhead.
 
Paulo Cereda
Paulo Cereda
7:56 AM
ooh rubber ducks
 
Skillmon
Skillmon
8:10 AM
@JosephWright \keyval_parse:NNn is not the equivalent of \setkeys, but I compare performance with \setkeys, so I think I should compare to \keys_set:nn (which is what I do).
 
Harald Hanche-Olsen
Harald Hanche-Olsen
Rubber duck debugging
In software engineering, rubber duck debugging is a method of debugging code. The name is a reference to a story in the book The Pragmatic Programmer in which a programmer would carry around a rubber duck and debug their code by forcing themselves to explain it, line-by-line, to the duck. Many other terms exist for this technique, often involving different inanimate objects. Many programmers have had the experience of explaining a problem to someone else, possibly even to someone who knows nothing about programming, and then hitting upon the solution in the process of explaining the problem. In...
 
Paulo Cereda
Paulo Cereda
@HaraldHanche-Olsen :)
 
Faheem Mitha
Faheem Mitha
8:39 AM
Oh, actually the \hrefs are arguments to \multirow. Perhaps that explains it.
 
Skillmon
Skillmon
@FaheemMitha not only perhaps. If you use it as an argument to anything, it wouldn't be able to change the category codes and fail.
 
Faheem Mitha
Faheem Mitha
@Skillmon Oh. Why can't it change the category codes?
 
Skillmon
Skillmon
@FaheemMitha once something was read by TeX, the category codes are fixed during the tokenization. If you now change the category codes this won't affect anything already tokenized. What hyperref does is, it first changes the category codes of problematic characters (like #) and after that reads its argument. If that argument hasn't yet been tokenized everything will work out correctly.
@FaheemMitha that's the same way \verb and the like work.
 
Faheem Mitha
Faheem Mitha
@Skillmon I see. So, a design feature?
 
Skillmon
Skillmon
@FaheemMitha a TeX limitation (or design feature of TeX, if you want to put it that way). Only way to get around that is using \scantokens, but that one has more pitfalls than it has benefits normally, so its usage has to be well considered.
 
Faheem Mitha
Faheem Mitha
8:46 AM
There is no way to do a first pass to see whether something inside that string wants to change the catcodes?
 
Skillmon
Skillmon
@FaheemMitha not without implementing a TeX parser in TeX. Not something you want to do (except if you're named Bruno)
2
 
Paulo Cereda
Paulo Cereda
@Skillmon THANK YOU
 
Faheem Mitha
Faheem Mitha
@Skillmon Fair enough.
 
Skillmon
Skillmon
@FaheemMitha but I think hyperref could get # and the like to work if nested, only % is a real problem, but that would slow things down a lot in the approach I have in mind, and has to be done really carefully.
 
Faheem Mitha
Faheem Mitha
@Skillmon What approach is that?
 
Skillmon
Skillmon
8:52 AM
@FaheemMitha doing something like \edef\myinput{\stringify #1\endmark} and \stringify being something that converts every character in #1 to category code 12 with \string, but \stringify would have to watch out for spaces, opening and closing braces, and some other stuff.
 
Faheem Mitha
Faheem Mitha
@Skillmon Sounds tricky.
 
Henri Menke
Henri Menke
9:08 AM
@PauloCereda Even more stars!!!
@PauloCereda I can't actually decide yet, whether I want to use lf or nnn.
@PauloCereda Oh, and do you know pass already? This has changed my life!
 
Paulo Cereda
Paulo Cereda
@HenriMenke :)
@HenriMenke I will try nnn in my laptop, sounds promising. :)
@HenriMenke ooh let me check
@HenriMenke it looks quite interesting!
 
Skillmon
Skillmon
@FaheemMitha I guess he might also be able to use \detokenize, this is like \verb but can't distinct between \foo {abc} and \foo{abc}, but that shouldn't be an issue here (I think....)
 
Henri Menke
Henri Menke
@PauloCereda You might also like this: github.com/pwmt/zathura
 
Paulo Cereda
Paulo Cereda
@HenriMenke ooh
 
Skillmon
Skillmon
@HenriMenke yay, my everyday PDF-viewer (except on Windows, there I use MuPDF)
 
Faheem Mitha
Faheem Mitha
9:17 AM
@Skillmon I think you mean "distinguish".
 
Skillmon
Skillmon
@FaheemMitha oh, stupid mistake :) Of course.
 
Paulo Cereda
Paulo Cereda
@Skillmon yay
Rabbits are very high-technical
 
David Carlisle
David Carlisle
@FaheemMitha ?
 
Faheem Mitha
Faheem Mitha
@DavidCarlisle Like something.
Which then becomes clickable.
 
David Carlisle
David Carlisle
@FaheemMitha isn't that what you asked for originally? perhaps I don't understand. A hyperlink whether its hyperref or html or whatever is always a link text and a specification of where to jump to if you activate the link isn't it? whether the markup is [foo]{bar} or <a href="bar">foo</a> or \hyperref{foo}{bar}
 
Faheem Mitha
Faheem Mitha
9:30 AM
@DavidCarlisle Well, one can just write the link as a URL, or one can write it in the form [text](URL).
I was asking about the latter. I already know that hyperref converts URLs to clickable links.
Though I thought the syntax was \href.
 
David Carlisle
David Carlisle
@FaheemMitha only if you specify the url as the link text as well as the url, \href{url} is just \hyperref{url}{url} it also makes Section 2 link to the section, and tables of contents link etc, most links generated by hyperref don't have the link target as the link text
 
Faheem Mitha
Faheem Mitha
@DavidCarlisle I'm not sure I follow. I'm doing \href{URL}{text}. This works. I didn't see \hyperref in the manual, but I didn't try to read all of it.
Oh, I see there are two different versions of \hyperref.
It doesn't look like I need to use those.
 
David Carlisle
David Carlisle
@FaheemMitha oh that's probably the form not looked at the manual for years I forget which is which, but the basic principle is that the fundamental command takes a link text and a target and any other link forms are just special cases of that
 
Faheem Mitha
Faheem Mitha
@DavidCarlisle The fundamental command being \hyperref?
 
Paulo Cereda
Paulo Cereda
9:53 AM
@DavidCarlisle quack
 
David Carlisle
David Carlisle
@FaheemMitha if that's the one with two arguments yes:-)
@PauloCereda Dinner
 
Paulo Cereda
Paulo Cereda
@DavidCarlisle oh no
 
Joseph Wright
Joseph Wright
10:14 AM
@DavidCarlisle, @UlrikeFischer, @egreg LaTeX2e 2020-02-02 PR3 and expl3 2020-01-19 off to CTAN
 
David Carlisle
David Carlisle
@JosephWright :-)
@JosephWright I had to update hhline, to keep Frank happy.
 
Joseph Wright
Joseph Wright
Meanwhile, any views on github.com/josephwright/siunitx/issues/371 would be useful
@DavidCarlisle Sure: I'll send tools in a few minutes: just re-building on Travis-CI
 
Paulo Cereda
Paulo Cereda
@JosephWright ooh a mouse
 
David Carlisle
David Carlisle
@JosephWright I know I've worked at NAG for 20+ years but at heart I'm still a pure mathematician, values are either wrong or right, never uncertain (and practically speaking I have no actual experience of typesetting these values other than occasional examples posted here)
 
Joseph Wright
Joseph Wright
@DavidCarlisle :)
@DavidCarlisle I think I'm going to push for NIST/BIPM to make their statements clearer, and talk to the chap about us deciding on a 'standard' by force of implementation
 
Faheem Mitha
Faheem Mitha
10:29 AM
@DavidCarlisle \hyperref takes both 2 and 4 arguments, according to the documentation. There are other commands in the hyperrefpackage that take two commands.
@DavidCarlisle Sometimes propositions are undecidable.
 
Joseph Wright
Joseph Wright
 
Paulo Cereda
Paulo Cereda
@JosephWright it's your fault
 
Joseph Wright
Joseph Wright
@DavidCarlisle Hope my reply on the XeTeX list doesn't conflict too much with yours
@egreg, @UlrikeFischer (@DavidCarlisle) We should I guess track down all the \tl_<thing>_case:n and \str_<thing>_case:n answers and alter to \text_<thing>case:n/\str_<thing>case:n
 
Paulo Cereda
Paulo Cereda
@DavidCarlisle engineers bah :)
 
Joseph Wright
Joseph Wright
@PauloCereda Probably
 
Paulo Cereda
Paulo Cereda
10:34 AM
@JosephWright :)
 
David Carlisle
David Carlisle
10:48 AM
@JosephWright no it's fine actually I'm glad you mentioned nfd I was trying to remember what was the additional information from the main unicodedata file, rather than just the smaller specialcasing etc
@FaheemMitha yes but you never write true(+/-0.5)
@JosephWright guess who I got mail from.
 
egreg
egreg
@JosephWright After the kernel is updated on CTAN.
 
Joseph Wright
Joseph Wright
@egreg Sure, yes, tomorrow or the day after I guess
@DavidCarlisle It's more recent but it's necessary (or at least I'd rather not have to hard-code the ~300 codepoints that are required for Greek), and we might expect over time more data to be added
@DavidCarlisle :)
@DavidCarlisle Er, Murray?
@DavidCarlisle The thing is, we might avoid reading UnicodeData.txt by pre-digesting, but for the common case of a format file that's not really necessary (The older XeTeX pre-digested stuff was not much faster than parsing everything on-the-fly)
 
David Carlisle
David Carlisle
11:04 AM
@JosephWright yes I think it should do format files, the pressure will be to push tikz (aka l3draw) hyperref some tagged pdf stuff plus all of expl3 into the format, then he has to load all that at run time on the simplest hello world document, it just seem wrong
@JosephWright you won't guess:-)
 
Paulo Cereda
Paulo Cereda
@DavidCarlisle ooh the Australian bloke
 
Joseph Wright
Joseph Wright
@DavidCarlisle I think even without that there's a big difference between a set up using plain and one using LaTeX; just the 'classical' kernel is already non-trivial
 
David Carlisle
David Carlisle
@JosephWright yes but he's timed that and decided it is workable to load every time but if by the time he's ready for release the kernel is rather bigger it may be better to have a backup plan
 
Joseph Wright
Joseph Wright
11:27 AM
@DavidCarlisle Yeah, perhaps, but the experience of others running service-like set ups is more toward custom formats
@DavidCarlisle Load time for core expl3 isn't actually so bad
 
Joseph Wright
Joseph Wright
11:41 AM
Benchmark #1: lualatex test.tex
  Time (mean ± σ):      7.875 s ±  1.381 s    [User: 4.4 ms, System: 15.1 ms]
  Range (min … max):    5.082 s … 10.287 s    10 runs

Benchmark #2: lualatex-dev test.tex
  Time (mean ± σ):      3.975 s ±  0.957 s    [User: 6.3 ms, System: 12.1 ms]
  Range (min … max):    3.061 s …  5.650 s    10 runs

Summary
  'lualatex-dev test.tex' ran
    1.98 ± 0.59 times faster than 'lualatex test.tex'
@PauloCereda ^^^
 
Paulo Cereda
Paulo Cereda
@JosephWright ooh :)
 
Joseph Wright
Joseph Wright
Benchmark #1: xelatex test.tex
  Time (mean ± σ):     10.947 s ±  1.062 s    [User: 10.2 ms, System: 18.8 ms]
  Range (min … max):    8.346 s … 12.208 s    10 runs

Benchmark #2: xelatex-dev test.tex
  Time (mean ± σ):      6.082 s ±  1.883 s    [User: 6.0 ms, System: 12.6 ms]
  Range (min … max):    4.812 s …  9.383 s    10 runs

  Warning: The first benchmarking run for this command was significantly slower than the rest (9.383 s). This could be caused by (filesystem) caches that were not filled until after the first run. You should consider using the '--warmup' option to fill those cache…
@PauloCereda (@DavidCarlisle) ^^^
@PauloCereda A useful tool
 
yo'
yo'
@JosephWright yaaaay!
 
David Carlisle
David Carlisle
@JosephWright which tool is that? Does show why (with no format) he wants to not read the unicode-data though:-)
 
Paulo Cereda
Paulo Cereda
@JosephWright we ducks are very good at finding tools. :)
 
Joseph Wright
Joseph Wright
11:53 AM
@yo' Of course, this is not going to scale: it's the load-time that's shorter
 
Paulo Cereda
Paulo Cereda
@DavidCarlisle I am useful
 
Joseph Wright
Joseph Wright
@yo' My test was loading ctex and printing 'hello world'
 
yo'
yo'
@JosephWright well, but that's the big annoyance: that a small document took ages to even start compiling.
 
David Carlisle
David Carlisle
@PauloCereda so you are
 
Joseph Wright
Joseph Wright
@yo' Yes, true: saves a lot of time on the test suite for example
@DavidCarlisle I did think about that ...
 
Paulo Cereda
Paulo Cereda
11:56 AM
@DavidCarlisle oh no
 
Paulo Cereda
Paulo Cereda
12:42 PM
It's too quiet today.
NOISEEEEEEEEEEEEEEEEE
ooh
 
Phelype Oleinik
Phelype Oleinik
Oct 7 '19 at 10:28, by Paulo Cereda
We Brazilians are very quiet. :)
:-)
 
Paulo Cereda
Paulo Cereda
@PhelypeOleinik oh no
@PhelypeOleinik shhh
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda WHAT DID YOU SAY!?
 
Paulo Cereda
Paulo Cereda
@PhelypeOleinik Hit the road Jack and don't you come back no more no more no more... ooh
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda Ooh, I heard this before, somewhere :)
@PauloCereda The song, I mean
 
Paulo Cereda
Paulo Cereda
12:49 PM
@PhelypeOleinik Tell me it was in the Super Mario Bros show that was broadcasted on TV when we were kids. :)
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda I don't know where. But I didn't watch that show. Was it good?
 
Paulo Cereda
Paulo Cereda
@PhelypeOleinik youtube.com/watch?v=qoLllXlNcmI (no PT-BR, sorry)
@PhelypeOleinik Around 12:40. That was the first time I heard Ray Charles. :)
@PhelypeOleinik well, being a child living in a rural area, I didn't have a lot of options. The show was good. :)
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda The live action guys in the beginning remind me of Castelo Ra-Tim-Bum
 
Paulo Cereda
Paulo Cereda
@PhelypeOleinik ooh que som é esse? :)
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda That show was good (most of the ones in TV Cultura were :)
 
Paulo Cereda
Paulo Cereda
12:56 PM
@PhelypeOleinik Yes. :) Remember mundo da lua?
TV Cultura started the daily broadcast by playing the national anthem. Good times.
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda I think so... vaguely :)
 
Paulo Cereda
Paulo Cereda
@PhelypeOleinik :)
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda I don't remember about that
 
Paulo Cereda
Paulo Cereda
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda What I remember is that from 5 to 6pm there was an hour of political propaganda. That hour took ages to end
 
Paulo Cereda
Paulo Cereda
12:59 PM
It starts around 0:22.
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda 2015? The video quality looks a tad older :-)
@PauloCereda late 1990s, maybe...
 
Paulo Cereda
Paulo Cereda
@PhelypeOleinik 1996 IIRC.
@PhelypeOleinik Dude, my eyes water with the national anthem, holy cow.
 
Phelype Oleinik
Phelype Oleinik
@PauloCereda It is a beautiful song, indeed :)
 
Paulo Cereda
Paulo Cereda
@PhelypeOleinik Yes!
 
yo'
yo'
1:50 PM
@JosephWright btw, we're discussing this in Overleaf just now, and we'll try to get TL2020 as fast as we can, especially for lualatex speeding up.
 
Joseph Wright
Joseph Wright
2:12 PM
@yo' Sure, but I know you need to test carefully there
 
yo'
yo'
@JosephWright yeah, we will. (Well, testing takes quite some time for TL19 as well just now)
 
Joseph Wright
Joseph Wright
2:54 PM
@yo' :)
 
Faheem Mitha
Faheem Mitha
3:34 PM
@yo' Overleaf has a lot of good/useful introductory TeX tutorials, and they seem to be of relatively recent vintage. Or Google very recently noticed they existed. I don't know which, but I've only started seeing them very recently. Do you know who wrote them?
 
Joseph Wright
Joseph Wright
@FaheemMitha There are Plans in the area of basic tutorials :)
 
Faheem Mitha
Faheem Mitha
@JosephWright Plans?
 
yo'
yo'
@FaheemMitha as far as I know, most are by my colleague LianTze Lim.
 
Joseph Wright
Joseph Wright
@FaheemMitha I have ideas about having a properly curated set of beginners 'lessons'
 
Faheem Mitha
Faheem Mitha
@yo' Ah. Well, congratulate him from me. How old is Overleaf?
 
Joseph Wright
Joseph Wright
3:36 PM
@yo' Ah, that's useful to know
 
yo'
yo'
@FaheemMitha her
@FaheemMitha about 8 years
@JosephWright what for?
 
Joseph Wright
Joseph Wright
@yo' Who wrote the Overleaf content
 
Faheem Mitha
Faheem Mitha
@JosephWright I see. You know, 'curated' might have been a reasonable word once, but is now hideous corporate buzzspeak.
 
yo'
yo'
@JosephWright yeah, but I wonder why is it useful to know the author :)
 
Faheem Mitha
Faheem Mitha
@yo' I beg your pardon. Her.
 
Joseph Wright
Joseph Wright
3:37 PM
@yo' :)
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle when I compile this 
\documentclass{article}
\usepackage{hyperref}
\begin{document}
\url{www.abc.de}
\end{document}
 
Paulo Cereda
Paulo Cereda
@JosephWright ooh that secret. :)
 
Ulrike Fischer
Ulrike Fischer
 
Alan Munn
Alan Munn
@DavidPurton Do you still need help with this? I have Acrobat XI Pro (on a Mac).
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle ^^^ It happens only with texlive, and the problem seems to be the ps2pdf step.
 
Faheem Mitha
Faheem Mitha
3:43 PM
@JosephWright Actually, there is already that LaTeX Wikibook thing - en.wikibooks.org/wiki/LaTeX
 
David Carlisle
David Carlisle
@FaheemMitha yes that's why we need to have some good tutorials
@UlrikeFischer ah what ghostscript do you have
 
Faheem Mitha
Faheem Mitha
@DavidCarlisle It's quite reasonable, and relatively useful. But it would help to develop it further.
 
David Carlisle
David Carlisle
$ gs --version
9.27
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle I think 9.50
 
David Carlisle
David Carlisle
@UlrikeFischer ^^
 
Faheem Mitha
Faheem Mitha
3:45 PM
@DavidCarlisle It's quite reasonable, and relatively useful. But it would help to develop it further.
Much of it is just stubs.
 
Alan Munn
Alan Munn
Site problems?
 
David Carlisle
David Carlisle
@AlanMunn not here, it seems
 
Alan Munn
Alan Munn
@DavidCarlisle Now I'm getting a site generated error page at least.
 
Paulo Cereda
Paulo Cereda
@AlanMunn normal behaviour here...
 
Alan Munn
Alan Munn
@PauloCereda Seems to have recovered by itself. Weird.
Must have been @UlrikeFischer 's fault.
 
Faheem Mitha
Faheem Mitha
3:51 PM
@AlanMunn Which site?
 
Alan Munn
Alan Munn
@FaheemMitha It was our site (TeX.se). Lasted a few minutes for me.
 
Faheem Mitha
Faheem Mitha
@AlanMunn Oh. I just had some problems with sending messages to chat. I thought it was my connection (which tends to be flakey). Possibly related.
Seems to have gone away now.
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle the ghostscript in texlive has version 9.50, in miktex it is 9.25. Was there an option somewhere to force texlive to use an external ghostscript?
 
Alan Munn
Alan Munn
@FaheemMitha Chat was working fine for me (which is usually the case; the chat and the regular site are independent from each other, AFAIK).
 
Faheem Mitha
Faheem Mitha
You know, what Linux needs is a nifty logo, like the TeX and LaTeX ones.
@AlanMunn Right. It might have been unrelated. I.e. my crappy connection.
Unfortunately, LiNuX looks kind of weird.
Meaning, a text logo. Not Tux.
 
David Carlisle
David Carlisle
4:04 PM
@UlrikeFischer I think it's always external except on windows
 
Paulo Cereda
Paulo Cereda
Works here with GS 9.27
 
David Carlisle
David Carlisle
what happens if you add -dNOSAFER
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle well it is clearly the border, with gs 9.50 I get a value of [0 0 8.33333302] instead of [0 0 1]. And if I suppress the hyperref border patch with \def\Hy@BorderArrayPatch{} the border behaves again <--@JosephWright
@DavidCarlisle doesn't change the output.
 
Faheem Mitha
Faheem Mitha
Can someone confirm that the unicode-math package loads fontspec?
 
David Carlisle
David Carlisle
4:21 PM
@FaheemMitha yes (as you can see in the terminal output)
 
Faheem Mitha
Faheem Mitha
@DavidCarlisle Oh. I didn't look at that. Sorry.
Thank you for the confirmation.
 
Skillmon
Skillmon
@JosephWright with category code safety against active commas and equals I'm currently at 3.5 times slower than keyval, but that should further increase, I just patched it in pretty ugly to test things.
 
Rmano
Rmano
4:42 PM
Hmmm... tex.stackexchange.com/questions/524076/… Strange. Seems some problem in finding the bounding box at \end{axis}
The MWE works in my standard Ubuntu distro but fails in an up-to-date TeXLive.
 
David Carlisle
David Carlisle
@Rmano there have been recent updates to tikz/pgf....
 
Rmano
Rmano
@DavidCarlisle Yes... I suggested the OP to report it.
Ah, ok, the Cat nailed it. But it's strange too; old circle(2pt) still works in tikzpictures, but fails in axis--- well, it has been deprecated for ages, so maybe it's all ok.
 
Alan Munn
Alan Munn
5:02 PM
From my university travel office:
I'll take the readable version please. :)
 
Paulo Cereda
Paulo Cereda
@AlanMunn ooh :)
 
Joseph Wright
Joseph Wright
@Skillmon Pretty impressive then
@Skillmon I've wondered about chasing more speed in l3keys, but it's tricky to find much and keep all of the features; in the end, if we want maximum performance then xtemplate's plan is the way to go - pre-parse the keys and use the actual values only at runtime
 
yo'
yo'
5:32 PM
@DavidCarlisle agreed.
 
Joseph Wright
Joseph Wright
@Skillmon I guess I could look to squeeze a bit more from l3keys ... I suspect \keyval_parse:NNn is as tight as I can make it, or very close to, but there might some wiggle room in the higher-level code
 
Pablo González L
Pablo González L
6:16 PM
 
Ulrike Fischer
Ulrike Fischer
@PabloGonzálezL ah, I knew there was something. But I wonder if it still works, at least locally in the command line it seems to have no effect.
 
Pablo González L
Pablo González L
@UlrikeFischer Win or Linux?
 
Ulrike Fischer
Ulrike Fischer
@PabloGonzálezL windows.
 
Pablo González L
Pablo González L
@UlrikeFischer With TeXLive 2018 it worked (win10) playing with the environment variables...
 
Ulrike Fischer
Ulrike Fischer
@PabloGonzálezL I will make a real try later, I made only a short test with set on the command line.
 
Pablo González L
Pablo González L
6:28 PM
@UlrikeFischer I used to test GS compatibility for a script that way (in the past), now I use Hyper with saved points for the different tests :)
 
Skillmon
Skillmon
6:57 PM
@JosephWright ok, now I'm down to 3.3 times slower than keyval with (from my few tests) robustness against active commas and equals, and without unpredictable brace stripping. With overall 238 lines of code.
 
Joseph Wright
Joseph Wright
@Skillmon Could you send me your test file?
 
 
2 hours later…
Ulrike Fischer
Ulrike Fischer
8:35 PM
@DavidCarlisle I think I found the change to the link border in ghostscript: git.ghostscript.com/…. And it probably means that we will need (and also l3backend) a variant "before gs 9.50" and one after.
 
David Carlisle
David Carlisle
@UlrikeFischer so they are intentionally not following the spec, to follow distiller, what happens if distiller gets fixed?
 
Sebastiano
Sebastiano
Good evening to all users active at this moment.
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle ;-(. But for us it means that we need some user switch to activate/deactivate this BorderArrayPatch, I don't think that we have any chance to detect the gs version.
 
Sebastiano
Sebastiano
I hope you're in good health and everything's okay.
 
David Carlisle
David Carlisle
8:54 PM
anyone want to bet against @egreg preparing an l3regex answer at the current time?
@UlrikeFischer yes I guess so, bit of a pain.... I wonder if we could make it part of the texlive (miktex?) site setup oh can't we write the test in postscript?
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle I don't quite see a site setup test if people like you can use more or less arbitrary external ghostscript versions. And I don't know postscript enough to decide if BorderArrayPatch can contain some "detection" code for the ghostscript version ;-(.
@DavidCarlisle no.
 
David Carlisle
David Carlisle
@UlrikeFischer well the PS test could be (GPL Ghostscript) product eq 950 revision gt and but I'll have to look what the patch does....
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle basically it is adding a factor to the border size. We discussed this last year: chat.stackexchange.com/transcript/message/50072059#50072059
 
David Carlisle
David Carlisle
@UlrikeFischer it's using ps code there so we could wrap it in {......} (GPL Ghostscript) product eq 950 revision lt if
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle If you can add it to hdvips.def I can test as I have both ghostscript types.
 
David Carlisle
David Carlisle
9:08 PM
@UlrikeFischer OK give me a sec...
 
Skillmon
Skillmon
@JosephWright I can do, it is currently not that well written but serves the purpose for me. What I still want to add is some automatic test suite for possible bugs (e.g., does it handle braces correctly, does it handle catcodes), those I do by hand currently.
 
David Carlisle
David Carlisle
@UlrikeFischer mail sent
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle got it. Just trying.
 
Skillmon
Skillmon
@JosephWright I'll tidy up a bit, expect mail in an hour (my wife will come home soon, so it might take a while)
 
David Carlisle
David Carlisle
@UlrikeFischer testing? it will never catch on.
 
Ulrike Fischer
Ulrike Fischer
9:29 PM
@DavidCarlisle it changed something - texlive still has a large border but in miktex it is gone completly now ;-( If I exchange the two /BorderArrayPatch is seems to work.
 
David Carlisle
David Carlisle
@UlrikeFischer I probably got the logic (or the arguments to ifelse reversed:-) let me check the manual
 
egreg
egreg
@DavidCarlisle I thought you knew about \check@mathfonts.
 
David Carlisle
David Carlisle
@egreg yes but it was middle of the night and I didn't look at the ams code I just stuck in the empty math list as it looked suspiciously like the kind of problem that would fix:-)
@egreg I left a comment under my answer
 
egreg
egreg
@DavidCarlisle :-)
 
David Carlisle
David Carlisle
@UlrikeFischer I think it should be 
    %if  using gs and release < 9.50 patch, otherwise make this a no-op
    (GPL Ghostscript) product eq revision 950 lt and
    {
    /BorderArrayPatch{%
      [exch{%
        dup dup type/integertype eq exch type/realtype eq or%
        {BPToDvips}if%
      }forall]%
    }def%
    }{
    /BorderArrayPatch{} def
    }
    ifelse
@UlrikeFischer I guess this is kind of urgent if windows releases (or anyone who has updated gs recently) gets massive borders in all hyperref docs by default
 
Ulrike Fischer
Ulrike Fischer
9:40 PM
@DavidCarlisle it doesn't work for me. I need this order: 
  %if  using gs and release >= 9.50 make this a no-op
    (GPL Ghostscript) product eq 950 revision le and
    {
    /BorderArrayPatch{} def
    }
    {
    /BorderArrayPatch{%
      [exch{%
        dup dup type/integertype eq exch type/realtype eq or%
        {BPToDvips}if%
      }forall]%
    }def%
    }
    ifelse
   % end of gs version test
@DavidCarlisle we would probably get a good overview about how many people still use the dvips route ;-)
 
Joseph Wright
Joseph Wright
@UlrikeFischer More than one ;)
@UlrikeFischer Lots and lots of publishers, Japanese users, ...
 
David Carlisle
David Carlisle
@UlrikeFischer i'm confused. your working version says "if ((gs=product) and (9.50 <= revison) do nothing else patch but don't you need the patch for old gs ??
@JosephWright you've got mail!
 
Joseph Wright
Joseph Wright
@DavidCarlisle Er, that's odd, I'll take a look
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle I clearly need the patch with miktex (which has an internal, older ghostscript). And it looks as if my version of the code enables the patch, but I don't understand the logic ...
 
Ulrike Fischer
Ulrike Fischer
10:07 PM
@DavidCarlisle does the miktex ghostscript count as "GPL Ghostscript "?
 
David Carlisle
David Carlisle
@UlrikeFischer ah can you run gs on the commandline (it may be called gswin32c or something similar, and type product == to the gs> prompt?
@UlrikeFischer like this session: 
$ gs
GPL Ghostscript 9.27 (2019-04-04)
Copyright (C) 2018 Artifex Software, Inc.  All rights reserved.
This software is supplied under the GNU AGPLv3 and comes with NO WARRANTY:
see the file COPYING for details.
GS>product ==
(GPL Ghostscript)
GS>
 
Ulrike Fischer
Ulrike Fischer
@DavidCarlisle for miktex I get this: 
GS>product ==
(MiKTeX GPL Ghostscript)
GS>revision ==
925
 
David Carlisle
David Carlisle
Grrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr
 
Ulrike Fischer
Ulrike Fischer
I can't do it for the internal texlive, it says it can't find gs_init.ps
 
David Carlisle
David Carlisle
@UlrikeFischer I'll look up what the ps for "contains substring" is:-)
 
Ulrike Fischer
Ulrike Fischer
10:17 PM
@DavidCarlisle ;-). I will go to bed now - skiing is tiring ...
 
David Carlisle
David Carlisle
@UlrikeFischer have a holiday, you don't have to do tex (I'm sure I have never done tex support or chatted here when on family holidays:-)
 
Skillmon
Skillmon
Did you know that TeX is faster in doing \catcode`\@=11 than in doing \catcode`\@11?
 
David Carlisle
David Carlisle
@Skillmon ald with a space after =?
 
Skillmon
Skillmon
@DavidCarlisle slower than without it.
 
egreg
egreg
@Skillmon Try 
\documentclass{article}
\usepackage{l3benchmark}

\ExplSyntaxOn
\cs_new_eq:NN \benchmark \benchmark:n
\ExplSyntaxOff

\begin{document}

\benchmark{\catcode`@=11 }

\benchmark{\catcode`\@=11 }

\benchmark{\catcode`@11 }

\benchmark{\catcode`\@11 }

\end{document}
On my machine I get 
7.41e-8 seconds (0.381 ops)
7.35e-8 seconds (0.374 ops)
8.06e-8 seconds (0.405 ops)
8.07e-8 seconds (0.408 ops)
 
Skillmon
Skillmon
10:31 PM
@egreg I'm getting similar values.
@JosephWright you got mail.
 
egreg
egreg
@Skillmon With the = it takes slightly less time because the optional equal hasn't to be looked for, otherwise TeX sees the 1, puts it aside and reprocesses it.
 
Skillmon
Skillmon
@egreg sounds logical.
 
Joseph Wright
Joseph Wright
@Skillmon Thanks
 
David Carlisle
David Carlisle
10:49 PM
@UlrikeFischer (when you have time) 
    %if  using gs and release < 9.50 patch, otherwise make this a no-op
    product (Ghostscript) search {pop pop pop  revision 950 lt } {pop false} ifelse
    {
    /BorderArrayPatch{%
      [exch{%
        dup dup type/integertype eq exch type/realtype eq or%
        {BPToDvips}if%
      }forall]%
    }def%
    }{
    /BorderArrayPatch{} def
    }
    ifelse
If anyone else has gs 9.50 and wants to try^^ :-)
@UlrikeFischer go away
 

« first day (3361 days earlier)      last day (1573 days later) »