@DavidCarlisle quack

hm

@marmot Thanks

@JouleV You're welcome.

@JouleV funny/ironic joke :)

@PauloCereda A matter of principles.

@yo' :)

@DavidCarlisle regarding the \MakeUppercase question for hyperref. Do you have any theory why Heiko choose the way to give an error?

@JosephWright Do you want me to add some doc to l3build?

@UlrikeFischer no, I suppose he thought just silently not uppercasing was bad. Could of course use Joseph's expandable l3 uppercase....

@DavidCarlisle If you want. I've just uploaded to fix the most urgent issue.

@DavidCarlisle if expl3 were in the kernel ;-). loading expl3 with hyperref could be an interesting experiment too ...

@JosephWright oh I didn't notice you'd pushed. I'll add something to the doc for next time, we should probably document the file use. Not sure if we need an option to rename the .curlopt file? otherwise need to document that you might want to add it to .gitignore etc

@DavidCarlisle Yes, would be good

@UlrikeFischer it could do l3 uppercase if defined else \@firstofone

@JosephWright OK I'll switch back to master and add something there.

@DavidCarlisle Great, thanks

@DavidCarlisle puh, complicated ifthenelse-stuff and perhaps with the need to do it at begin document.

@UlrikeFischer I was thinking if doing it at run time ie define the bookmark uppercase to be \ifx\whatever\expandafter\@firstofone\else\expandafter\l3_uppercase\fi{#1}

@JosephWright @UlrikeFischer did you see Bruno objected to my latex3 command names?

@DavidCarlisle and he also adjusted the logic, the first now has two true cases ;-)

@DavidCarlisle I see Bruno's not so keen on my expansion plans

@JosephWright I really don't see a problem if you go ahead - users of l3draw and users of old engines are non-overlapping user groups.

@UlrikeFischer Same here

@UlrikeFischer I'm going to do some of the other tweaks first, leaving f-type in place, and see if the email thread evolues

@UlrikeFischer @JosephWright I added some words to l3build.dtx, hope they make sense, my typing is rarely right first time...

@DavidCarlisle looks good. Is the "in" correct?: "It can also be useful to check in the |curlopts| file ..."

@UlrikeFischer I'll change to "check the contents of ...

A palindrome even in the world of @DavidCarlisle :-)

@marmot ↓↓ Smuggles as far as possible within the confines of a TikZ environment ↓↓

\documentclass{article}
\usepackage{tikz}

\makeatletter
\newcount\smugglecnt
\newcommand*\smugglewithintikz[1]{% %% <- smuggles as far as possible without leaving TikZ
    \advance\smugglecnt 1 %% <- only incremented locally to reduce macro usage (jfbu's idea)
    \expandafter\global\expandafter\let\csname smuggle@\the\smugglecnt\endcsname#1% %% <- store dfn
    \aftergroup\ifx\aftergroup\scope\aftergroup\tikz@scope@env %% <- test if we're in a TikZ env.
      \aftergroup\let\aftergroup#1\expandafter\aftergroup\csname smuggle@\the\smugglecnt\endcsname %% <- smuggle

@marmot: ↓↓ Smuggles up to (optionally including) the depth of a target marker ↓↓

\documentclass{article}

\makeatletter
\newcount\smugglecnt
\newcommand\smuggleone[1]{%
  \advance\smugglecnt 1  %% <- only incremented locally to reduce macro usage (jfbu's idea)
  \expandafter\global\expandafter\let\csname smuggle@\the\smugglecnt\endcsname#1%
  \aftergroup\let\aftergroup#1\expandafter\aftergroup\csname smuggle@\the\smugglecnt\endcsname
}
\newcommand*\smuggletarget[1][]{% %% <- creates a target for smuggling operations
  \expandafter\edef\csname smuggle@target@#1\endcsname{\the\currentgrouplevel}%

I'm sure they can both still be improved.

@Circumscribe Oh WOW! (Now I feel guilty, pinging you in the chat was just to gauge whether or not this is possible, not to ask you to work that out.) That's great!! (May I ask an official question such that these things get shared with the community?)

@marmot It was fun, so there's no need to feel guilty.

@Circumscribe OK, thanks! (Should I ask an official question?)

@marmot Hmmm… (that's the sound of deliberation)

I should mention btw that \smuggleuptotarget smuggles up to (and including) the depth of the nearest \smuggletarget with a matching optional argument.

@Circumscribe In the past, smuggling was a crime. And now it is a talent. ;-)

@marmot: I do wonder what kind of goods you're smuggling and if I should be endorsing this :)

@Circumscribe Only honey liquor and so far we stayed in the legal limits.

@marmot It's not smuggling if its legal :)

@Circumscribe I know, that's a big problem. That's why I am not very good at it and have to ask questions how to properly smuggle. ;-)

@marmot Fair enough :)

@ChristianHupfer Your trauma is eased, congratulations!

@DavidCarlisle hello sir :-)

@touhami hi

@DavidCarlisle i was just passing, good day for you :-)

@marmot This does not look like honey liquor to me!

Those look like numbe… wait… are you using it just to count how many items there are in a list? O_O

@DavidCarlisle Yes, there is finally meaning of life for me again ;-)

@Circumscribe Yes, is there an easier way to count them (without loading additional packages, with packages it is easy)?

@marmot: I'm now pondering this. I think just using a global macro with a sufficiently unique name would probably be better.

@Circumscribe Why would that be better? I thought if one confines the stuff to the local group this might be better. There is already enough harm coming from the fact that some users do \foreach \x in .... while \x and \y have its own meaning in the calc syntax.

@marmot: \smuggle also defines \smuggle@<number> globally, and it now defines a new one for every iteration.

@Circumscribe Yes, but this does not interfere with other commands. Your counter makes sure that these things are unique. On the other hand, if I stored the number of items in a global macro, and the user would have two of them, they would get overwritten. I understand that in this case it is fine but I still do not feel comfortable doing this.

@marmot: I agree that it would be better to not define (non-unique) anything outside the current scope. Doing it this way just feels a bit like overkill.

@Circumscribe Yes, it would be nice if TeX had a builtin feature like "broadcast in current environment" that would not require to define any global macros. Still I think your smuggling is a great thing to have, at least all the stupid macros like \myangle can be made local in tikz pictures.

@marmot Or variables that can be set in the scope where they were originally defined from within a narrower scope.

@Circumscribe Yes, something like this. TikZ does a lot of trickery with lengths like \pgf@xa which are recycled all over to keep the damage minimal.

@marmot: Maybe it'd be good to define a command \getlistlength{#1}{#2} that just assigns the length of list #1 to #2.

@marmot: and uses getlistlength@dummy in the way you describe.

@Circumscribe Yes, I agree. I feel that there could be tons of helpers like that which might define the content of a tikz library or something like this. But then others will object "why don't you use xparse or something, why do you reinvent the wheel?" You have a username that allows you to deflect such arguments easily, but I don't. ;-)

@marmot: Well, you can always answer "because I'm a marmot" :)

@Circumscribe That's a good point. ;-) (No, I really mean that for lists there is xparse, listofitems and you know not, so these things exist in principle. But if you load them, this may cause issues, and the answer may not be as appreciated as an answer that does not rely on these things.)

@marmot: Yeah, I see your point. I also don't really like telling someone to load a package if it's only needed once for one simple thing that you could do differently.

@marmot: My original remark wasn't meant as criticism by the way. The O_O was more about how TeX sometimes makes seemingly simple things complicated.

Hi I use biber and biblatex and want the bibliography with a comma or colon between author and title like: A. Einstein: "About Bosons and Bisons" or A. Einstein, "About Bosons and Bisons" where can I configure the , or :

Now I have a "." there. I searched in the biblatex doc, but I have no clue what to search for.

@marmot: ↓↓ Look what I found :D ↓↓

\documentclass{article}
\usepackage{tikz}
\begin{document}
\foreach \X [count=\Y,remember=\Y as \numthings (initially 0)] in {M,A,R,M,O,T} {}
\numthings
\end{document}

I'm actually somewhat surprised that this works, because the documentation (p976) does not describe this behaviour. (Also, "(initially 0)" was redundant.)

The scope of \numthings is precisely what you would like it to be though.

@JonasStein What style are you using?

@AlanMunn I use style=numeric

:48974438 \documentclass{article}
\usepackage[style=ext-numeric]{biblatex}
\DeclareDelimFormat[bib,biblist]{nametitledelim}{\addcolon\space}
\addbibresource{yourbibfile.bib}
\begin{document}
\cite{yourcitekey}
\printbibliography
\end{document}

Documentation in biblatex-ext.

@JonasStein ^^^

@AlanMunn works perfect. Thank you!

Do you think I should write a question for this so that others can learn from it too? I think it was not asked before. Or is that out of scope?

@JonasStein We actually do have a question (I couldn't find it though, so just gave you a solution.) tex.stackexchange.com/q/103815/2693

@JonasStein Unfortunately the accepted answer is out of date, but moewe's answer shows the current way to do this. And actually it seems you don't really need to change the style to ext-numeric to change this particular delimiter.

@AndréC Hey, you're right! I had just blindly assumed \Y would no longer be defined outside the loop (like \X).

@Circumscribe Yes, looks great!

@marmot : @AndréC's version is better.

@Circumscribe Yes, looks like it. (In these loops one never really knows at which level things happen. \X is local, \Y is not, good to know.)

@AlanMunn This is a problem with tex.sx: A good answer in 2011 can be complete nonsense in 2019.

@JonasStein Yes, true, but since new answers can be added, and comments added to old answers (or in some cases old answers edited) it's less of a problem than, e.g. mailing list archives.

users are not active anymore and the solution marker is never set to the new, better answer

@JonasStein Yes, the acceptance issue is trickier. Since despite the official rhetoric, we do tend to think of the accepted answer as the "best/most correct" even though it's really just the one that helped the OP at the time.

@JonasStein In more obscure questions like this one, the problem is worse, since the new answer is unlikely to surpass the vote count of the older answer.

@JonasStein Although I think for regular users of the site, it doesn't take long to realize that reading all the answers to a question is the best strategy.

Yes. But anyway you helped me a lot now, and I upvoted the linked Q/A

@JonasStein So did I. :) Glad I could help.

@MarcelKrüger would it be possible (without to much fuss ...) to have an option "don't stretch spaces"? I do find the new behaviour much better, but as it changes existing documents ...

@AlanMunn This seems to suggest that one has to rethink the definition of a duplicate. If an answer got the tick at that time because it solved the problem then, I feel it would be a bit unfortunate if the tick was removed because a new answer appeared which uses a package that was written later. So this is a bit tricky, and it might be worthwhile to allow for a new question being asked without immediately closing it as a duplicate. Or maybe not.

@marmot I don't see why it's a big problem. Questioners are always free to change their accepted answer when another one comes along. The problem is more that old accepted answers hang around. A comment + a new answer should be sufficient to direct users to the new answer. And if the new answer gets accepted later, even better, IMO. The tiny loss of reputation to the original accepted answerer is neither here nor there.

@AlanMunn It is not the reputation loss but the fact that a new question is very visible. Let's say a question has been asked, and answer written and accepted. Later there is a new package allowing for a better solution. Now you would just post an answer to the old question. I am afraid that it won't be very visible. And then precisely what you describe above will happen: users will see the highly voted answer, which has a good chance to remain the accepted answer.

@marmot The visibility of the question in a search is much more related to total number of views, and other factors such as how well it describes the problem. I do agree that there are cases where new questions are entirely appropriate, but as a general principle, i.e., make a new question instead of adding a new answer to an old question, I don't think it's necessary.

@AlanMunn I agree that it depends on the case. It is just that not many go through old questions and check if by now a better answer could be written. But we are very efficient at closing questions as duplicates (which is in principle a good thing) but maybe not all of us check whether the answer to the original question is still state of the art. In fact, if I am not mistaken, the only criterion for being able to close something as a duplicate is that the original Q has an upvoted A.

@marmot Yes, that's sometimes a problem. And the oldest questions on the site have generally worse answers (even from good people) than later questions, simply because people were learning how to write them. But a duplicate can be pointed in either direction (i.e., you can add a new answer and then close the old question as a duplicate.)

@AlanMunn Does that happen often?

@AlanMunn Hmm, I am not a fan of closing an earlier Q as a dupe of a later one.

@Circumscribe I've done it a couple of times. Every so often we also get a really frequent question and then it's helpful to write a canonical answer. This happened e.g., here: tex.stackexchange.com/q/140814/2693. It probably should happen more often.

@marmot The main duplicate should be the one with the best question/answer combination. Age of the question is irrelevant, IMO.

@AlanMunn Well, often "being a duplicate" is associated with "you have not searched carefully enough to realize that this has already been asked before".

@marmot no that's the wrong approach, closing as duplicate should not be seen as a criticism of the person asking the question.

@marmot I agree with @DavidCarlisle. And anyway, in the case of closing an older question pointing to a newer question, it has no effect on the older OP, since (i) they're probably long gone (most users are transient) and (ii) it happens usually a long time after they ask the question.

@DavidCarlisle But this is a bit odd, isn't it? Assume I have a question. Shouldn't I try to find out first whether or not it has been asked before? If yes, I upvote it and the answer. Are you suggesting: no, just write a new question, and if you happen to get an answer that is superior to the answer the earlier question got, the earlier question will be closed as a duplicate?

@AlanMunn @DavidCarlisle I am really confused. Should users just ask new questions, not look at earlier posts, and hope that things have changed in such a way that a simpler answer might emerge?

@marmot No, that's not what I mean. If the old question answers things well, then yes, upvote it and move on. But this conversation started by talking about outdated answers and whether new questions would be appropriate. But sometimes, someone asks the question in a better way than the older one, and receives a better answer, and later we realize there's a duplicate. In the case, the best question/answer pair should be the one that is the target of the duplicate link, not the older one.

@marmot in some cases people should look but often its hard, when I am searching for duplicates to close questions I usually search for the answer but that's not an option the questioner presumably has. Especially if i know I have posted an answer before, it's a lot easier to find. so in almost all cases the newer question gets closed straight away, but if retrospectively we find two posts that could be duplicate I would not consider the age of the post top be relevant

@Circumscribe Do you mean questions with a duplicate link already, or simply two questions which are duplicates of each other and unlinked in either direction?

@AlanMunn Questions that are already linked, but in the wrong direction.

@AlanMunn @DavidCarlisle IMHO users should try to first find related posts, and if they are not happy with the solutions presented there they should post a new question, saying they have read X but do not like Y and want Z instead.

@Circumscribe note if they are really duplicate (that is the answer to one question answers the other) you can flag them and ask they be merged, but usually we use "duplicate" in a weaker sense that it's the same underlying issue and anyone reading one ought to be able to figure out th eother

@Circumscribe Probably the easiest is to raise them here in chat pinging one of the mods. They can reverse the links if appropriate.

@AlanMunn Alright, that sounds reasonably simple.

@AlanMunn @DavidCarlisle So if I understand you correctly, A being duplicate of B means that A is inferior to B (and both have a common answer). Is that how you view this?

@DavidCarlisle I didn't know that was possible; good to know.

@marmot Sure, in an ideal world. But as @DavidCarlisle said, it's not always easy to find them. I'm just repeating David here, but most times I find duplicates by searching for what I know will be in the answer. This is clearly not a strategy that questioners can use.

@marmot Yes, if both questions have answers. Obviously no, if one question has no answer. And crucially it's the question that must be the same, not the answer.

@AlanMunn Yes, but it suits the purpose: you provide the OP with a link to an answer.

@marmot partially although it's actually a mess because often a much clearer question gets added later which we close as duplicate just because reversing the links to make the new question have the answers is just too much work for not much gain. I think the model is that there is no implied superiority of the question chosen as the non closed post, it's just one of the set and having lots of duplicates of it increases its search profile and makes it easier to find

@AlanMunn Yes, that's what I meant. (It is not clear what it means to be the same question, let's say A wants a line to become red and B it to become blue, then I would say they have a common answer but these are technically different questions.)

@marmot Sure. The extreme case of this is How to look up a symbol. But sometimes people ask entirely different questions that so happen to have the same solution, and people try to close as a duplicate. This is bad.

@DavidCarlisle Yes, this is how it is used. In the best of all worlds things would all merge to one "original" question with a set of answers, and the other questions, which have the same content but different wording, will be linked as duplicates.

Well it looks as if letterspacing in luaotfload broke ;-(

@AlanMunn I actually do not see the problem. Let's say we have two questions which sound very different but have a common answer. Why would it be bad to link them in that way?

@marmot I vary my opinion on this, for many questions of the form "why have I got spurious space" get closed by the canonical "what is the use of percent at end of lines" question. That is sort of OK as anyone finding the question is directed to a usable answer, but the question is not at all duplicate one is questioning the symptom, the other is questioning the technique used in the answer there is no way someone struggling with a space from {{<newline> is going to find the % question

@marmot Pointing out that the other question exists is probably good. Closing as a duplicate is bad, since questioners search for the problem not the answer. Of course we could find individual cases (like your blue/red example) but I think in most cases are not like that.

@DavidCarlisle @AlanMunn Very good. Now I have you where I wanted you to be. ;-) My starting statement was that we may have to rethink our practice of closing questions as duplicates. What you are saying now seems to be consistent with that. And yes, I agree with your last statements.

@marmot I think hoping for consistency is unrealistic, I'm not even consistent myself, it works well enough in practice

@AlanMunn @DavidCarlisle Of course the statement was not 100% serious but I agree. ;-) In particular with the assertion that the current situation is a mess (but wouldn't know how to improve it...)

@DavidCarlisle To much exposure to LaTeX? ;-)

@marmot Really? This is a different issue. You started out saying that that old questions might benefit from new questions with updated answers rather than the new questions being closed as duplicates. But nothing @DavidCarlisle and I have said here really agrees with that. We may agree that there are some issues with duplicates on the site, but that's cherry-picked version of your original statement. :)

@AlanMunn "Closing as a duplicate is bad, since questioners search for the problem not the answer."