TeX, LaTeX and Friends - 2018-04-27

« first day (2735 days earlier) ← previous day next day → last day (2181 days later) »

1:30 AM

Is the TeX Live TUG mailing list moderated? I subscribed, at least for the time being, and sent a message (a possible bug report) there, but it hasn't shown up in the archives yet.

2:19 AM

@RandomDSdevel -- the message landed in my mailbox 15-20 minutes ago. i'm not sure that messages are always sent to the sender; some lists don't. (but it's been so long since i was added to that list that i don't remember.)

2:51 AM

Stack Overflow Isn’t Very Welcoming. It’s Time for That to Change.

Jay Hanlon on April 26, 2018

Let’s start with the painful truth:

Too many people experience Stack Overflow¹ as a hostile or elitist place, especially newer coders, women, people of color, and others in marginalized groups.

Our employees and community have cared about this for a long time, but we’ve struggled to talk about it publicly or to sufficiently prioritize it in recent years. And results matter more than intentions.

Now, that’s not because most Stack Overflow contributors are hostile jerks. The majority of them are generous and kind. Sure, a few are… just generous, I guess? But our active users regularly express thei …

2 hours later…

4:50 AM

@s.patroller I don't think it's the same for TeX.SE: the just-do-it-for-me questions not only are answered but also upvoted! And I have never read any sexist or racist comment. However, also here, rules are elastic, depending on if you part of the "elite" or not.

5:09 AM

@CarLaTeX I agree, and that is something I like with TeX.SE (the first part).

5:31 AM

@mickep :)

1 hour later…

6:36 AM

@CarLaTeX this "elite," "clan" idea appears to be a common theme:

in The Classroom, 7 hours ago, by Buffy

@s.patroller My first experience with trying to contribute to SO was very disappointing. It seemed to reek of testosterone. It isn't welcoming of newbies at least and there is a lot of impatience with people who aren't in the clan.

@s.patroller Curiously, I didn't experience it here when I was a newbie, it happened to me after some time.

7:08 AM

@CarLaTeX ~~@egreg~~ people were (and still are) very rude to me, since I started using the site :(

7:25 AM

@DavidCarlisle Perhaps because you're mean and eat ducks :P

7:53 AM

Jun 29 '17 at 16:15, by Paulo Cereda

@DavidCarlisle you are not mean :)

May 30 '17 at 10:03, by Paulo Cereda

@CarLaTeX You are mean!

@DavidCarlisle Now I'm waiting for @HaraldHanche-Olsen reply about déjà-vu...

@DavidCarlisle Oh no, I don't have a "you're not mean" message!

@CarLaTeX of course not, but you have at least three "you are mean" ones. Seems about right!

@DavidCarlisle :'(

@CarLaTeX You mean one like:

Harald Hanche-Olsen

@CarLaTeX Who, me? I'm much too nice for that:

Apr 9 at 14:40, by Paulo Cereda

@HaraldHanche-Olsen: you are not mean :)

8:05 AM

@CarLaTeX You are not mean ;)

2

Harald Hanche-Olsen

@CarLaTeX Indeed, the different stackexchange sites seem to have developed widely different cultures. This one seems better than most, but then I don't have a good nose for detecting abuse unless it becomes extreme.

@DavidCarlisle @CarLaTeX @PauloCereda I seem to make neither of the two, which also seems about right!

I'm tempted to say: Dear SO, we (at TeX.SE) have been telling you for ages!

4

Feb 15 '14 at 13:21, by Paulo Cereda

@tohecz You are mean. I love you. :)

@DavidCarlisle now that's an old one!

@TeXnician <3<3<3

@HaraldHanche-Olsen LOL

@HaraldHanche-Olsen Perhaps we don't know English enough to consider "sweetie" an offence...

8:20 AM

@yo' Funny though: we can't be sure we get it right either ...

@JosephWright sure thing, but we can sign under basically all points made there!

@yo' You're not mean <3 (preventive)

@CarLaTeX well, I got the "I love you" in my "You are mean", so I'm happily living, mean and loved :-) (almost like in Stronghold PC strategy game, where you could be "Cruel and worshipped"!)

@s.patroller Getting this right is tricky, I agree: there's a complex balance between letting new people ask what they need to know and avoiding the issues forums have: 'support fatigue'

4

Harald Hanche-Olsen

@CarLaTeX Hmm. I think I might, though I sometimes read too fast and miss such things. I have noted that in the UK, people of either gender tend to address those of the opposite gender as luv', or does that go in only one direction? Not sure how to interpret it, though.

8:23 AM

@HaraldHanche-Olsen I'd be wary of using "luv'": it feels to me like something that would be said only by a man to a woman, and unless they know each other well would be somewhat 'loaded'

@CarLaTeX Another one I'd avoid

Harald Hanche-Olsen

@JosephWright Somewhat related – I find it really difficult to find duplicates, even in cases when I am virtually certain there must be several. Not sure what can be done about it, though. But it makes it harder to yell at people for either asking or answering questions when there are duplicates.

4

@HaraldHanche-Olsen Yes, true

Harald Hanche-Olsen

@JosephWright Indeed, it seems infantilising.

@CarLaTeX in a work place if you call male developers "developers" and female developers "girls" or "sweetie" or pretty much anything that isn't the same as you'd call their male colleagues, it's not good even though the words themselves in other contexts are not offensive

6

@HaraldHanche-Olsen A lot of questions come down to the same core issues, but it's not always clear that this will help a questionner. I favour sorting in comments or as an answer, then duping if that works

8:26 AM

@HaraldHanche-Olsen this is an issue, but it's much more difficult than it seems to laymen.

@DavidCarlisle oh yes, the old "developers" and "coffee machines" issue

@yo' Finding duplicates is itself tricky, though perhaps that means we need to be more aggressive in editing 'canonical' questions to make them easy to find

@JosephWright not sure about that. But I think that there should be an option in the search to privilege search results with many dupes. Maybe the close-as-dupe form search does this, I dunno, the thing is that you don't use this one to look for the dupe, right?

@yo' Also a reasonable point

@yo' It varies: we've got a few canonical questions that have been posted deliberately and are very clear. On the other hand, there are some I know I've duped to that are a bit vague.

@JosephWright there could be a badge for "vote to dupe and edit the dupe" :-)

Harald Hanche-Olsen

@DavidCarlisle I think I know of only one case where “sweetie” is part of the job title: That's “Miss Sweetie Poo”, usually a 12-year old girl filling an important rôle in the Ig Nobel prize ceremony.

8:41 AM

@HaraldHanche-Olsen where were you in the UK? I was at my parents last week (Nottinghamshire) and it was again noticeable how you always got called luv or duck or some other general term of endearment by shop assistants and other complete strangers. Around here (Oxfordshire) that basically never happens:-)

Harald Hanche-Olsen

@DavidCarlisle I don't recall exactly where, really. It's been too long ago. But I don't specifically recall hearing it in Oxford, now that you mention it. Southampton, perhaps, maybe Glasgow.

8:57 AM

@CarLaTeX Well. Sometimes. Sometimes they're downvoted and closed.

@JosephWright good point

9:34 AM

Hey

I have a problem with PStricks (barcodes specifically) not rendering, and the TeX log shows absolutely nothing (it compiles without error - just stops when the barcode would hit and everything after that is blank)

@user5389107 We'll need some form of demo input

@JosephWright pastebin.com/0nr7dWmy

No error, just crashes the ghostscript with errorcode 1 while compiling

This is pdfTeX, Version 3.14159265-2.6-1.40.17 (MiKTeX 2.9 64-bit) (preloaded format=latex 2016.10.4)

9:49 AM

@user5389107 I meant a full example ...

Just putting that in a doc and trying to compile triggers the error

The full document is 22k lines long I doubt you'd want that

@user5389107 No, but you should be able to cut it down to a few lines

Alright, might take a while

need to exclusion test which of the ~90 packages cause this

I think it's because the data doesn't fit into a 26x26 datamatrix code. But it did with the old ms access add-in

10:16 AM

Quack!

Friday, yay!

10:26 AM

@yo' You didn't see the not

@DavidCarlisle Yes, of course, in Italian is the same

@TorbjørnT. Rarely :):):)

10:45 AM

One issue I have is that \begin{pspicture}(18mm,18mm) my picture ignores the format specifiers

It extends way beyond 18mm

It seems to ignore format specifiers

@user5389107 it is like a standard \begin{picture} the argument says how big an area latex should leave, but the picture itself can overprint the whole page it is not cropped or scaled to that size

11:13 AM

@DavidCarlisle Noticed that

Somehow I think I'm not specifying the parameters right, but the docs to pst-barcodes are cryptic and it has no errorhandling

So I'm completly in the dark here

11:32 AM

And whenever something doesn't work, PSTricks just crashes the ghostscript compiler outright without providing any indication of what is happening

@user5389107 in general that is unavoidable, it is generating postscript code as a string which it can not interpret and that is merged with the postscript that dvips generates from the dvi and if the resulting combination has postscript errors then this will not be detected until ghostscript sees the file

Any way to get the exact error there or do I just need to live with no error handling?

I don't know which option or combination of options is failing because nothing tells me that. I need to know why my barcode isn't working without trial and erroring hundreds of combinations.

@user5389107 if you use gs on the commandline it should report any errors. (I haven't used pstricks for years so this is all a somewhat distant memory)

Part of my problem stems from the complexity of the toolchain

@DavidCarlisle unicode-math has interesting side-effects. Did you saw github.com/wspr/unicode-math/issues/458?

11:46 AM

The compiling is handled by a php class that assembles the actual tech code from a header template, footer template and body template with dynamic sub-report templates multiplied and filled in. It then runs it through the MikTeX toolchain and I only get access to the latex log and ghostscript log.

@user5389107 doesn't the ghostscript log have sensible error messages?

MiKTeX GPL Ghostscript 9.19: Unrecoverable error, exit code 1

In this case

Thats all I get

@user5389107 you must be calling it with commandline flag -q or an API equivalent to suppress logging I guess

Maybe, I don't have access to the server or code that does that

@user5389107 well if someone is throwing away the debug information it's hard to debug....

11:51 AM

indeed

@user5389107 it should be like this, if I write abc.ps with some bad PS:

(xy)

1 1 add
pop

pop

kji

and run gs then it says:

$ gs abc.ps
GPL Ghostscript 9.21 (2017-03-16)
Copyright (C) 2017 Artifex Software, Inc.  All rights reserved.
This software comes with NO WARRANTY: see the file PUBLIC for details.
Error: /undefined in kji
Operand stack:

Execution stack:
   %interp_exit   .runexec2   --nostringval--   --nostringval--   --nostringval--   2   %stopped_push   --nostringval--   --nostringval--   --nostringval--   false   1   %stopped_push   1999   1   3   %oparray_pop   1998   1   3   %oparray_pop   1982   1   3   %oparray_pop   1868   1   3   %oparray_pop   --nostringval--   %errorexec_pop   .runexec2   --nos…

@user5389107 you seem to only have the last line, which is not much use.

I won't be able to turn it on

The dev with access to do that is on vacation longer then I have time left

@UlrikeFischer yes although the "hidden math" in url handling is pretty strange thing anyway...

12:15 PM

ctan.org/pkg/glossaries provides a download glossaries.tds.zip which is always overwritten with the latest version. This is not ideal for distribution via Gentoo Linux. Is there a download source for tex packages with version numbers? Sth. like glossaries-4.12.zip glossaries-4.37.zip ...

Anytime they squeeze the duck, it says kwack.

@PauloCereda ^^

@yo' OH

12:37 PM

@JonasStein not really (but don't you take texlive rather than taking things from ctan???) ctan only has the latest versions, texlive of course has svn source control so in theory you can get whatever.

12:47 PM

How I can have both vertically centered text in a cell AND automatically calculated with of cell AND longtable=

@Maïeul tabu or ltxtable or ltablex give you X columns in longtable, and m columns give you vertical centering

@HaraldHanche-Olsen -- finding duplicates can be really hard when what you're looking for is something like a double backslash or a code segment that isn't all letters. that's one reason why some of us (at least me) try to add really good (and sometimes not quite so good) questions to the "often referenced questions" list. maybe it would help (in searching) to make that list more prominent to newbies.

X columns that is centered ? because I tried with it, and I get problem with merged cells

@DavidCarlisle -- well, i do refer to the three recent additions to the group i work with as "well scrubbed kids", but i wouldn't do that to their faces. (but i think i might be forgiven, because their cumulative ages are less than mine. and they're really good guys -- smart, competent and generally courteous without becoming sappy.)

@HaraldHanche-Olsen -- "miss sweetie poo" is, by definition, eight years old. and she's a real trouper! (my husband and i have been attending the ig nobel ceremonies for a lot of years now. always the "nth first annual".)

I don't understand how I should use X type of columns

1:02 PM

@Maïeul yes in tabularx itself you need to redefine \tabularxcolumn from p{#1} to m I think tabu has some interface for setting that on the column

@DavidCarlisle for now, I d'ont understand what X type of columns does

Harald Hanche-Olsen

@barbarabeeton Oh, I remembered the age wrong, then. Yes, eight makes more sense for that than twelve, I suppose. Wish I could attend myself, but the travel is just too much. I sometimes watch the live stream, however, or else the video afterwards.

@Maïeul sorry not much time now but it calculates the column widths (see tabularx package manual)

ok

@JosephWright we should get @egreg using travis, seen this feature:

1:04 PM

I thouh it was for vertically centered

user image

@DavidCarlisle Anathema!

@egreg I thought you would be pleased

arf

that does not work

maybe because I sue rotatebox inside some of my column head

1:25 PM

@Maïeul it is difficult to sort out alignment in tabular without concrete code.

@UlrikeFischer yes, but my concrete code has some personnal commands that I need to explode to show you

@DavidCarlisle yes ;-). I'm rather certain that at least with luatex there must be something better which doesn't use math.

@U

@UlrikeFischer @DavidCarlisle I found a way to get what I need using parbox

1:44 PM

user image

@JosephWright @egreg ^^

With \expanded?

@egreg no that's just checking the build builds before we break anything, so just a matter of checking some check boxes in the travis setup (which is how come I saw the new travis feature:-)

@DavidCarlisle Two can play at that game ...

1 hour later…

2:51 PM

@DavidCarlisle that sounds interesting. I just had a look through their servers and found tug.org/historic/systems/texlive/2016/tlnet-final/archive This could be perfect for us. The server is very slow, but we mirror everything, so it is no problem. Thank you a lot. I will discuss it in our tex-Team

3:36 PM

@JonasStein I don't know the context but installing one of those should be a rather rare and specialist activity, they are frozen snapshots in time with no update paths other than uninstalling it and installing a later one, unlike a current texlive where you can update individual packages vi tlmgr

1 hour later…

4:46 PM

@PauloCereda Do you remember the TeXprinter issue I sent you some while ago? I've found the cause.

5:11 PM

@TeXnician I remember your issue, and I am in debt with a lot of people recently... :(

@PauloCereda No worries, when I've got a fix (or something more), I'll poke you ;)

@TeXnician ooh poking

Dr. Manuel Kuehner

5:34 PM

@DavidCarlisle Hi. Short update regarding tex.stackexchange.com/questions/425632 (you helped me a lot there):

@DavidCarlisle
I contacted the author of KOMA (Markus Kohm, Germany) and here's summary of what he recommended.

Better use \sbox instead of \setbox.
Or use \color@begingroup and \color@endgroup when using \setbox.
The definition of \sbox shows how it should/could be done.
In the case of paracol (mad box tricks that span multiple pages) you need to take care of color stack.
I also contacted the author of the paracol package (Hiroshi Nakashima, Japan). He wants to look into it.

I also contacted Heike Oberdiek, the author of color-stack-related packages. He won't be able to look into it in …

@DavidCarlisle The formatting got lost. I also put the update in the question (tex.stackexchange.com/questions/425632).

5:46 PM

Old LaTeX 2.09 files can still be compiled on current TeX Live systems, right?

I haven't tried in a while.

@FaheemMitha I think so.

@egreg the wrath of David breaking stuff. :)

@PauloCereda Maybe 2.09 documents are not affected by UTF-8 woes. :-)

@egreg Exactly. :) What has � ever done for us?

3

6:07 PM

@PauloCereda You mean apart from bringing us aqueducts? (or aqueduckts? ;-)

@marmot Quack. :)

6:25 PM

@DavidCarlisle The problem is, that Linux distributions have difficulties to migrate the tlmgr into their packagemanagement system. Finally we pull the source from ctan like in the package packages.gentoo.org/packages/dev-tex/glossaries

6:43 PM

@egreg Ok. Thank you.

@JonasStein yes I realised that you used the system package manager rather than tlmgr but I thought you pulled from the texlive repositories rather than pulling from ctan and redoing all the work of deciding what goes where (for packages without a tds zip on ctan)

Personally I use tlmgr, but I help in the Gentoo-TeX team and we use ctan sources mostly. It is bad that ctan changes a file without changing its name. This breaks our packages often.

a frozen file from the texlive ftp would be a good alternative. At the moment we do not provide the daily update tlmgr provides

@JonasStein don't get me wrong, but this sounds like you are blaming CTAN for their choosing of package maintenance...

we just provide update ~once a year, when the next texlive release is out

@Dr.ManuelKuehner I wonder who added those color@begingroups to \sbox :-) If you find an issue with any of the obediek bundle packages leave an issue on github and it mostly comes to me:-)

6:48 PM

I do not blame them, because I do not pay them. But it is correct that it would make our work easier, if they would provide ams-3.45.tgz, ams-3.46.tgz and ams-3.47.tgz and probably ams-latest.tgz as link to the latest

In a ideal world we would manage to migrate the tlmgr into our packagemanager.

I do not know any distribution who managed this.

I am still a learner. I do not know the packaging and tlmgr internals well.

@JonasStein I don't see how you can say that is bad? the latex sources have been updated dozens of times on ctan over the last thirty years without changing the filenames, ctan is not a source code repository it is a distribution network and has the current files only. I'm sorry but it sounds like you are using it in a way for which it was simply not designed

@JonasStein that's unfortunate (for example you will get latex 2018-04-01 patch level 2 in texlive 2018 but you really want patch level 3)

@DavidCarlisle yes. Our package management expects that we get a different tar.gz per version. The proper solution would be to understand tlmgr by heart and migrate it. But I fear no distribution has the manpower to do so. As a result we get this bad solution.

@egreg that's a serious point actually, we could \UseRawInputEncoding in \documentstyle

It works well for python. Most distributions support frequent updates for pip/pypi packages, but not for ctan packages

@JonasStein does gentoo do similar with other language package managers, do python users not use pip or node users not use npm ?

6:55 PM

Well, it works fine with python and perl. For GNU-R it should work, but due to lack of manpower it does not work.

and with TeX... we have a chaos and try to tidy it up with 3 developers in sparetime.

Originally we had texlive packages like texlive-extra and additionally the same package as single package like amstex.

This caused a lot of problems, but we tidied up a lot.

I think, if it would be easy to migrate the tlmgr to a distribution, it would be already available in any other distribution with more manpower in the TeXteam

@JonasStein to be honest I can't see why the linux distributions don't just make the initial texlive installation a system package but let it be updated via tlmgr. There seem to be very few advantages to customising things (just looking from outside, and seeing repeated questions here with people on various linux distributions stuck on old or inconsistent tex installations)

I do not know it completely too, because I am still investigating and learning. But I found out so far, that some tex packages rely on external tools like perl

and we can not let tlmgr manage the TeX files, if all other dependencies will break

Some users want to use tools, which are used outside of texlive, but also installed via texlive.

@DavidCarlisle That would be a very technically difficult thing to do.

Distributions will have to provide these tools as individual package.

And messy.

Assuming it was even possible in a reasonable way.

7:04 PM

@FaheemMitha why? surely other languages update via their own update systems don't they?

Lets take minted package. This shows how many problems can occur.

I would love to learn, why it works for CPAN, but not so well for CTAN packages.

@DavidCarlisle Some software has its own subsystem to manage local package installs of add-on packages. I.e. R. But it makes arrangements to keep those separate from regular binary system packages. And in that case, R and its main libraries are still regular binary system packages.

That's probably the best you can do, aside from a purely local install.

When I say "make arrangements", it was probably the Debian R maintainer who made those arrangements.

I suppose it might be possible to do something similar for TeX Live.

I think it should be easy, if there were only *.sty and README

@JonasStein minted has a dependency on python pygments but I can't see why you can't just say that a user has to use the python(or linux) update system to install that. It seems odd to have to re-implement an entire tex package installation mechanism just because a couple of packages have dependencies on other programs

Dr. Manuel Kuehner

@DavidCarlisle Haha :). It's is very often you!

7:23 PM

@JonasStein well as a windows user I have to install tools like python on my own too, with miktex I even have to install perl. Imho if we can manage linux user should be able to handle this too.

Linux users are used to the package management and that they do not have to care about any dependencies. A fully working TeX installation should be one command on every distribution and it should be kept up to date in a user friendly way too.

Perhaps the solution would be simple but no linux developer has ever read the tlmgr code.

@DavidCarlisle It came to my mind while I was wlking home that there can be things such as \catcode`é=\active\def é{\'e}’

@JonasStein You could browse the TL archive list and gather information of everything related to what you or the team wanted or envision to achieve, then you can have a better background on the historical/technical reasons.

tug.org/historic/systems/texlive/2016/tlnet-final/archive << you mean this one?

Yes this is a good help. I will not fix TeX on Linux over night. It may take years, until we will have the daily updates on Linux as we want to have.

have to go, see you later

7:50 PM

@egreg yes that's probably the main use for \UseRawEncoding

@JonasStein no tug.org isn't designed for lots of people downloading it (I seem to recall it cuts off around 25 connections) the live texlive packages as pulled by tlmgr (tl2017 now, soon tl2018) are on ctan and so mirrored all over so if you pull from mirrors.ctan.org it will find somewhere faster and closer than tug.org

8:20 PM

I'm a bit stuck.

I defined the following:

\newcommand{foobar}[1]{lots of text, including #1}

But I get errors with \foobar{foo}, that

! Missing number, treated as zero.
<to be read again>
                   o

thing.

I defined the macro in the body. Is that ok?

@FaheemMitha \newcommand{\foobar}[1]{lots of text, including #1}

@PauloCereda Oh, balls.

@FaheemMitha :)

Thanks. I'm kicking myself right now.

@FaheemMitha don't do that, it's normal. :) I once added margins=2m instead of 2cm. :)

8:26 PM

Well, as long as it doesn't happen too often...

@FaheemMitha :D

Why does that cause the "Missing number, treated as zero." error?

8:37 PM

@FaheemMitha at the place where it breaks the code is looking for the number of arguments but gets the o from foobar.

9:17 PM

@UlrikeFischer I suppose we could spare a couple of bytes to give a sensible error message these days...

2

@UlrikeFischer oh. Thank you.

Maybe one day TeX will actually produce useful error messages.

@FaheemMitha it's hard to get good errors from macro processing languages, and especially hard in latex2e as we had no spare tokens

@DavidCarlisle No spare tokens?

Does that make a difference?

@FaheemMitha in 1993 if you loaded latex2e+amsmath into emtex on an ibm pc you had around 50 spaces left in the string pool so 50 \newcommand or \label etc for the entire document and only just enough token memory to process a page, so if we used a couple of internal macros and made a better error check in newcommand that would be a 4% decrease in the number of commands a user could make in documents and package code

@DavidCarlisle Is it possible to write a better error check for newcommand than the existing one?

Question: is there an easy way to add an Attn: line to a scrlttr2 class? And is this interesting enough to post to the site? Before subject, normally.

9:26 PM

@FaheemMitha oh sure but changing a core definition like \newcomand always has a cost someone will be using \patchcmnd to patch \newcommand for some feature, and so changing the internal structure in any way will break something, that is always the problem with any change

@DavidCarlisle Oh.

2 hours later…

11:21 PM

@DavidCarlisle Couldn't whoever it is just repatch the thing?

« first day (2735 days earlier) ← previous day next day → last day (2181 days later) »

Transcript for

Apr '1827

$TeX - LaTeX$ TeX, LaTeX and Friends

\begin{chat}

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

$TeX - LaTeX$

site design / logo © 2024 Stack Exchange Inc; legal

mobile