« first day (3611 days earlier)      last day (1305 days later) » 

3:58 AM
I am trying to include a web link inside \begin{thebibliography}, but the URL is not properly displayed
I have used url = {}
also used url, hyperref package
 
4:31 AM
@PabloGonzálezL I am not sure what is happening, but according to the TeX Directory Structure, context will not look for flies which are under the tex/latex sub-tree of the TDS. So, I guess s l3backend-luatex.def should be located under tex/generic rather than tex/latex. (This is just a guess).
 
 
3 hours later…
7:32 AM
@Aditya @PabloGonzálezL the l3backend files where always there, and if it previously found l3backend-pdfmode.def it should find l3backend-luatex.def now there too. But the files where renamed some days ago (pdfmode was splitted in pdftex and luatex), so my guess would be that something is wrong with the filename database.
 
7:45 AM
@UlrikeFischer DeviceN fully-working: just have to think about export
@UlrikeFischer And adding tests :)
 
@JosephWright oh, I will be away the whole day, so can only look in the evening. I have been pondering about this blend color problem in beamer, but I don't see a good solution. If you want a global blend color traversing groups you can't really know how much of the color stack will be affected.
 
 
1 hour later…
9:14 AM
Is it possible that \unless is buggy? Consider the following "working" code
\makeatletter
\long\def\@secondofthree#1#2#3{#2}
\newcommand\myifheadspace[1]
  {%
    \if
        \iffalse{\fi\myifheadspace@\@empty#1? }%
        \relax\relax
      \expandafter\@secondofthree
    \fi
    \@secondoftwo
  }
\long\def\myifheadspace@#1
  {%
    \if\relax\detokenize\expandafter{#1}\relax\else f\fi
    \expandafter\@gobble\expandafter{\iffalse}\fi
  }
\makeatother
\expandafter\myifheadspace\expandafter{\detokenize{&#}}\TRUE\FALSE
This will correctly throw an undefined control sequence \FALSE error.
On the other hand the following will throw a missing \begin{document} error:
\makeatletter
\long\def\@secondofthree#1#2#3{#2}
\newcommand\myifheadspace[1]
  {%
    \if
        \iffalse{\fi\myifheadspace@\@empty#1? }%
        \relax\relax
      \expandafter\@secondofthree
    \fi
    \@secondoftwo
  }
\long\def\myifheadspace@#1
  {%
    \unless\if\relax\detokenize\expandafter{#1}\relax f\fi
    \expandafter\@gobble\expandafter{\iffalse}\fi
  }
\makeatother
\expandafter\myifheadspace\expandafter{\detokenize{&#}}\TRUE\FALSE
Even though it should be equivalent.
Oh, looks like the issue is that the remainder of the stuff after \if isn't gobbled correctly as #1 isn't necessarily a single token, leading to the first \if's test to be done early. My bad.
 
9:30 AM
ooh rabbit skills <3
 
@PauloCereda well, not really. It took me embarrassingly long to realise this :)
 
user image
4
@Skillmonlikestopanswers.xyz ^^ rabbits are good at coding :)
 
 
3 hours later…
12:24 PM
@PhelypeOleinik oh, I didn't consider the problem being \long and \protected... Will try.
 
@Skillmonlikestopanswers.xyz I wrote some of the debug code. I know what atrocities it can do :-)
@Skillmonlikestopanswers.xyz It's basically
\let \__debug_macro:Nx \macro:Nx
<prefixes from \macro:Nx> \def \macro:Nx #1#2
  {
    <debug checks>
    \__debug_macro:Nx #1 {#2}
  }
 
@PhelypeOleinik before yesterday I didn't ever take a look at l3debug, who needs debug code anyways if one doesn't make coding errors to begin with.
 
@Skillmonlikestopanswers.xyz So the \macro:Nx must have the right prefixes and signature
@Skillmonlikestopanswers.xyz That's a good point ;-)
 
@PhelypeOleinik take a look above :)
 
@Skillmonlikestopanswers.xyz Oh, the \unless\if<more than two tokens> thing. Been there, done that :-)
 
12:32 PM
@PhelypeOleinik and rabbits using their iCarrot laptops :)
 
@PhelypeOleinik Well, it is somewhat comforting that Bruno didn't see it at first look when he suggested something like that in the first place :)
 
@PauloCereda I'd use an iCarrot. Sounds funny :-)
 
@PhelypeOleinik ooh
 
@Skillmonlikestopanswers.xyz TeX's rules for conditionals are mind-twisting indeed. If I remember right, I've been where you are and wrote that optimisation for \prg_return_false: \else: \prg_return_true:, but Bruno convinced me to drop it (there were only one or two functions in the kernel that would use it, I think)
 
@PhelypeOleinik Maybe if we also get @JosephWright on our side we can convince Bruno :)
 
12:39 PM
@Skillmonlikestopanswers.xyz You never know ;)
 
@Skillmonlikestopanswers.xyz A mutiny?
 
@Skillmonlikestopanswers.xyz Frank is always keen on optimisation if we can find it safely
 
@JosephWright Well, tweaking \prg_new_conditional a bit to also optimize in the \prg_return_false: \else: \prg_return_true: \fi: case isn't hard at all, and twisting the tokens inserted to not need \exp_after:wN but only use intelligent argument grabbing and reinsertion instead doesn't seem unsafe at all.
 
@JosephWright premature optimization is the root of all evil
6
 
@Skillmonlikestopanswers.xyz We've had a few optimisations of the conditionals over time; we likely need to work through the cases that concern BLF
@Skillmonlikestopanswers.xyz Certainly speedups are good: next release will have some for LuaTeX
 
12:42 PM
@JosephWright yeah I loosely followed that one.
 
@Skillmonlikestopanswers.xyz Cool
 
@JosephWright and the first half of that was the reason why I started speeding up l3tl :)
@JosephWright I'm silently watching everything
 
@Skillmonlikestopanswers.xyz :)
@Skillmonlikestopanswers.xyz You want to be careful; that's how you end up reading all of expl3
 
@Skillmonlikestopanswers.xyz ooh bunny watchers
 
@JosephWright well, I often skim through stuff (e.g., I didn't read everything Marcel did in those Lua speed ups, more like recognised he does something)
 
1:00 PM
@PhelypeOleinik thanks for taking a look at the l3debug problem, it should now be correct.
 
@Aditya I got out of the way using 'context --generate' and everything worked again OK
@PhelypeOleinik You have received an email
Does anyone here know how to configure $TEXMFAUXTREES in bash...
I have tried export TEXMFAUXTREES='/path/folder:' but it doesn't work for me :(
 
@PabloGonzálezL it's just an environment variable (if you use the enviornment variable rather than texmf.cnf
 
@DavidCarlisle Yes, the thing is that if I use 'texmf.cnf' I break ConText (tug.org/pipermail/tex-live/2020-September/046099.html)
 
1:15 PM
@PabloGonzálezL that should work (you don't need the quotes unless there are spaces) depending where you set it you may also need to export it
 
@DavidCarlisle Unfortunately it doesn't work for me :(
 
@PabloGonzálezL can't say I have ever found a need for the auxtrees thing, surely if using kpathsea it shouldn't make any difference whetehr you use a texmf.cnf or the environment the same effective value will be used by latex and context won't it?
 
@DavidCarlisle Karl wrote a reply, but, he also doesn't know what is going on and will continue to use export TEXMFAUXTREES
@DavidCarlisle I know that ConText doesn't use 'kpathsea' (I've read the TeX Live documentation :), but, I don't know how to get out of it
 
@PabloGonzálezL i'm surprised you say that texmf.cnf is for advanced users, especially for windows users setting enviornment variables is somewhat alien and texmf.cnf was introduced specifically to give a cross platform uniform way to configure such thinngs
what does your package do that needs an entire new tree? can't you just set TEXINPUTS ?
 
@DavidCarlisle tlmgr --help conf
    Warning: The general facility for changing configuration values is
    here, but tinkering with settings in this way is strongly
    discouraged. Again, no error checking on either keys or values is
    done, so any sort of breakage is possible.
@DavidCarlisle They are a couple of collaborative work projects, I assume they used 'auxtree' when they modified everything to use TeX Live in exchange for MikTeX
 
1:30 PM
@PabloGonzálezL I wouldn't use tlmgr just edit the text file /usr/local/texlive/2020/texmf.cnf which says:
% (Public domain.)
% This texmf.cnf file should contain only your personal changes from the
% original texmf.cnf (for example, as chosen in the installer).
%
% That is, if you need to make changes to texmf.cnf, put your custom
% settings in this file, which is .../texlive/YYYY/texmf.cnf, rather than
% the distributed file (which is .../texlive/YYYY/texmf-dist/web2c/texmf.cnf).
% And include *only* your changed values, not a copy of the whole thing!
%
setting values there is no harder (in fact it is easier) and no more dangerous than setting them in the environment
 
@DavidCarlisle I think that's why the project uses texmf.cnf (everything comes ready to be configured by a script), but, in my case this breaks ConText, maybe I should add other variables inside my texmf.cnf, the rest of the project participants don't use ConText so they don't have problems.
 
@PabloGonzálezL Why do you try to set it with a var instead of using tlmgr conf?
 
@Skillmonlikestopanswers.xyz Looks good now
 
@UlrikeFischer I'm just trying alternatives, using 'tlmgr conf' everything is OK, except ConText which stops working
@PhelypeOleinik Hola, un lindo día para pasar en Git :)
 
2:01 PM
@PabloGonzálezL Replied :-)
@PabloGonzálezL Sometimes I regret learning git (not really :-)
 
2:30 PM
@PabloGonzálezL does it stops working because of the setting itself, or because of something in the new tree?
 
@UlrikeFischer Apparently by using a custom path to texmf.cnf this stops ConText, so I was trying to configure $TEXMFAUXTREES using environment variables
@PhelypeOleinik I am working on the answer :)
 
What does Bad conditionals: <num> in the log mean?
(well I guess that there were some bad conditionals, but I have no idea how this could've been triggered in this case...)
 
@PabloGonzálezL you mean you did set the TEXMFCNF variable? Yes context doesn't like this, but using texmf conf auxtrees doesn't set this.
 
Ah, ok it is triggered by regression-test.tex if at \END the \currentiflevel is bigger than 2. Though I have no idea how those ifs could still be open...
 
2:47 PM
But using only `tlmgr conf auxtrees --conffile /home/pablo/texmf/texmf.cnf add ~/path/folder/` without setting $TEXMFCNF doesn't work (I don't have
administrator permissions on that computer, only one configuration per account)
 
@PabloGonzálezL I never used conffile, I always use simply tlmgr conf auxtrees add.
 
@UlrikeFischer I should have access to 'sudo' and this is not the case :(
 
3:48 PM
@PabloGonzálezL I can't see how that can make any difference, if context sees the wrong path via texmf.cnf it will see the wrong path via env variable
 
@DavidCarlisle Mmm, you're right, I'm stuck in this, I don't see a way to make ConText not get lost in routes and let me use 'axutree'.
 
@PabloGonzálezL you can always just alias context to a command that unsets the variable. But why need a tree, if all the files are for that project can't you just set TEXINPUTS=.//:
 
@DavidCarlisle Unfortunately this is out of my control, I get out of the way by copying the files to my current directory, but it's tedious to do this every time they change something
 
@PabloGonzálezL surely your own shell alias is in your control? you can make context be a command that unsets whatever variable you want to unset then calls the normal one
 
4:08 PM
@DavidCarlisle Of course, it's an option, just like using TEXINPUTS, the strange thing is that I have to do these vicissitudes, the texmf.cnf file as you told me, is designed for these situations
 
4:34 PM
@PabloGonzálezL thinking about it, your path for the texmf.cnf looks odd. Why don't you put it in web2c in your texmfhome or texmf-local?
 
4:46 PM
@UlrikeFischer OK, but that file can be anywhere. Can you give me an example of how it should look like inside my texmfhome ?
 
@JosephWright do you happen to know when \tl_if_head_eq_charcode:nN and \tl_if_head_eq_catcode:nN were written? Because they work with the current optimisation done by \prg_new_conditional just by chance it seems. They will fail in the current implementation with the p type if I'm not mistaken (will further look into it later today, have to leave for now).
 
5:04 PM
@PabloGonzálezL if you use a texmfcnf variable, yes. But texmf.cnf has a standard search path like all other files, and if you put it in one of this place you won't have to set texmfcnf. My local texmf.cnf is in texmf-local/web2c.
 
5:18 PM
@PabloGonzálezL as Ulrike says and also you can simply put one in the directory of your project if you only want to affect the paths for that project
 
5:35 PM
I manage to copy the files, this is the full description of what happens :

$ echo $TEXMFCNF
$ pwd
/home/pablo/ltxcole
$ tree stycole/
stycole/
└── tex
└── latex
├── colegio-new2019.sty
├── logoaurora.jpg
└── tkz-custom.sty
2 directories, 3 files
$ tlmgr conf auxtrees --conffile /home/pablo/texmf/texmf.cnf show
tlmgr: no auxiliary texmf trees defined.
$ tlmgr conf auxtrees --conffile /home/pablo/texmf/texmf.cnf add ~/ltxcole/stycole/
$ tlmgr conf auxtrees --conffile /home/pablo/texmf/texmf.cnf show
 
so if it upsets context why do export TEXMFCNF='~/texmf:' why not just put the texmf.cnf file in a standard place?
 
@DavidCarlisle What would be the standard route? (without having to use 'sudo', only on my account)
If I don't put export I can't find .sty, if I put export I lose ConText :(
 
@PabloGonzálezL I don't think I understand the problem to be honest you could stick it in /usr/local/texlive/2020/texmf-local/web2c or in /home/pablo/ltxcole or just globally have a TEXINpUTS of .//: and not mess with the texmfcfg at all or...
 
5:54 PM
@DavidCarlisle The problem is that I don't have write permissions on /usr/local/texlive/2020/texmf-local/web2c (it's a computer shared by many users) in general I get out of the way with TEXINPUTS or copying the files temporarily under ~/texmf, but it forms a mess to keep order with git...I think I'll leave ConText aside and use it only on my personal computer :)
 
@PabloGonzálezL you could put a cnf file in /home/pablo/ltxcole presumably
 
@DavidCarlisle I tried several locations, but if I don't use export, the line kpsewhich colegio-new2019.sty returns nothing :_(
 
cis
@DavidCarlisle I was able to solve it ---> tex.stackexchange.com/a/563355/46023
 
6:18 PM
@PabloGonzálezL do you have any local texmf tree that works out of the box? What for example is the value of texmfhome?
 
$ kpsewhich -var-value=TEXMFHOME
/home/pablo/texmf
$ tree -d texmf
texmf
├── doc
│   └── latex
│   └── pgfornament
└── tex
├── generic
│   └── pgfornament
│   ├── am
│   ├── pgfhan
│   └── vectorian
└── latex
├── alterqcm
├── pgfornament
└── tkz-doc

13 directories
 
@cis oh ok, so you were asking to answer a question? I thought it was your data:-)
 
@PabloGonzálezL what do you get with kpsewhich -show-path=web2c? Is there a path mentioned for which you have write access?
 
$ kpsewhich -show-path=web2c
/home/pablo/.texlive2020/texmf-config/web2c:/home/pablo/.texlive2020/texmf-var/web2c:/home/pablo/texmf/web2c:!!/usr/local/texlive/texmf-local/web2c:!!/usr/local/texlive/2020/texmf-config/web2c:!!/usr/local/texlive/2020/texmf-var/web2c:!!/usr/local/texlive/2020/texmf-dist/web2c
 
6:34 PM
@PabloGonzálezL well then try one of them: put a texmf.cnf in the folder and look if it is found.
^^^ today in venice
 
@UlrikeFischer :(
$ cp ~/ltxcole/texmf.cnf ~/texmf/web2c/
$ kpsewhich -all texmf.cnf
/usr/local/texlive/2020/texmf.cnf
/usr/local/texlive/2020/texmf-dist/web2c/texmf.cnf
$ cp texmf.cnf /home/pablo/.texlive2020/texmf-var/web2c/
$ kpsewhich -all texmf.cnf
/usr/local/texlive/2020/texmf.cnf
/usr/local/texlive/2020/texmf-dist/web2c/texmf.cnf
 
@PabloGonzálezL but you did remove the texmfcnf settings did you?
 
I always come back to the same ones, I need to use 'export' and there I break ConText
@UlrikeFischer Yes, I have deleted the rest of the copies I have of texmf.cnf
 
@PabloGonzálezL no, you are right, the cnf path is not web2c. What do you get with kpsewhich --show-path=cnf? Is there a usable path?
 
$ kpsewhich --show-path=cnf
/usr/local/texlive/2020/bin/x86_64-linux:/usr/local/texlive/2020/bin/x86_64-linux/share/texmf-local/web2c:/usr/local/texlive/2020/bin/x86_64-linux/share/texmf-dist/web2c:/usr/local/texlive/2020/bin/x86_64-linux/share/texmf/web2c:/usr/local/texlive/2020/bin/x86_64-linux/texmf-local/web2c:/usr/local/texlive/2020/bin/x86_64-linux/texmf-dist/web2c:/usr/local/texlive/2020/bin/x86_64-linux/texmf/web2c:/usr/local/texlive/2020/bin:/usr/local/texlive/2020/bin/share/texmf-local/web2c:/usr/local/texlive/2020/bin/share/texmf-dist/web2c:/usr/local/texlive/2020/bin/share/texmf
A long list of directories in which I cannot write :(
 
6:56 PM
@PabloGonzálezL yes looks as if you are out of look. I guess if you put the texmf.cnf in the standard context folder where context looks anyway it explodes too?
 
@UlrikeFischer The only place I see something related to ConText is `.texlive2020/texmf-var/luatex-cache/context
And it's certainly not the right place
 
@PabloGonzálezL no, I mean where you standalone context looks for its configuration. perhaps texmf-context/web2c
 
When ConText works OK, this is the line you are looking for:
mkiv lua stats > used config file: selfautoparent:/texmfcnf.lua;selfautoparent:/texmf-dist/web2c/texmfcnf.lua

When I add 'export' it fails and throws:
resolvers | resolving | looking for regular 'texmfcnf.lua' on weird path '$HOME/texmf:' from specification '$HOME/texmf:'
resolvers | resolving | looking for fallback 'contextcnf.lua' on weird path '$HOME/texmf:' from specification '$HOME/texmf:'
 
7:13 PM
@PabloGonzálezL and what happens if you set texmfcnf to the context path whatever/texmf-dist/web2c/ ? Does context continues to work?
 
7:25 PM
@UlrikeFischer Yes, ConText stops working...now that I look at the 'textfcnf.lua' file a little bit, it contains the default paths
 
@PabloGonzálezL perhaps you should ask on the context list, if there is a value for texmfcnf which does not confuse context.
 
@UlrikeFischer I made the query in the TeX Live list, but I may have to make it in the ConText list. In the 'texmfcnd.lua' file, I read the following:
        -- The following variable is predefined (but can be overloaded) and in
        -- most cases you can leve this one untouched. The built-in definition
        -- permits relocation of the tree.
        --
        -- TEXMFCNF     = "{selfautodir:,selfautoparent:}{,{/share,}/texmf{-local,}/web2c}"
 
7:45 PM
Sometimes this latex3 test suite is killing me. Upon running l3build check in l3kernel the format for pdfTeX works like a charm, but the XeTeX format seems to enter an infinite loop.
 
@Skillmonlikestopanswers.xyz ;-). Is it already pushed?
 
@UlrikeFischer no, I always run at least the tests of l3kernel before pushing...
@UlrikeFischer error was created due to an undefined control sequence. I have no idea why I didn't get the error thrown right at me but instead got the XeTeX format building looping.
 
@Skillmonlikestopanswers.xyz that is a problem of scrollmode, that it often gets stuck only much later.
 
@UlrikeFischer is that an issue unique to XeTeX?
@UlrikeFischer and what is scroll mode?
 
@Skillmonlikestopanswers.xyz well \scrollmode, when latex doesn't stop at errors. You can get very far with it and then suddenly something explodes.
 
8:09 PM
@UlrikeFischer ahhh, now I get what you meant. I thought that was \batchmode.
 
@Skillmonlikestopanswers.xyz which ever, one of these commands which I never use ;-)
 
@UlrikeFischer :) Yeah, I didn't even know \scrollmode was a thing.
 

« first day (3611 days earlier)      last day (1305 days later) »