is there a way to \begin{macro}{} an \ifabc@stuff (i.e. a tex one) in ltxdoc? or do i just leave conditionals out?

@cfr does it not just work? (I coudl test it but...)

@cfr zref package for example has

zref/zref.dtx:%    \begin{macro}{\ifZREF@found}
zref/zref.dtx:%    \begin{macro}{\ifZREF@immediate}
zref/zref.dtx:%    \begin{macro}{\ifZREF@pl@list}
zref/zref.dtx:%    \begin{macro}{\ifZREF@pa@list}
zref/zref.dtx:%    \begin{macro}{\ifzref@titleref@expand}
zref/zref.dtx:%    \begin{macro}{\ifzref@titleref@stripperiod}
zref/zref.dtx:%    \begin{macro}{\ifZREF@xr@zreflabel}

@daleif \NewExpandableDocumentCommand?

@Skillmon ^^ this

@daleif As @Skillmon says

@daleif I used \fp_eval:n because \ang might not be integer.

@egreg of course, as always!

@Skillmon Both failed for me, but it was not that macro that failed, the OP had missed the ; at the end.

Did anyone here ever have a reason to play with \linepenalty? I get a feeling the effect of increasing it is minimal.

\linepenalty=10000?

@mickep Well, that's because most paragraphs within the specific line width have like at most +-1 on the line count. (And most paragraphs have basically 1 good line count and just a handful of good breakings.)

@yo' In a book I just saw a handful of changed paragraphs, and only one of them resulted in a shorter paragraph. (with linepenalty set to 100 or so)

@mickep no it's sort of like \looseness (which also only works sometimes) but the effect of looseness is a lot easier to understand so I'd always try that first

@DavidCarlisle Indeed. I wonder how Knuth came to the value 10 (in older TeX it was 1...)

@mickep I just tried with +/- 5000 and + made no difference and - changed the line breaking but not the number of lines (just stretched the space so the last line was longer which suggests changing this at all doesn't make much difference unless you have very stretch interword space so lots of flexibility in line breaks

@DavidCarlisle Indeed, same conclusion here. I guess the choice of 10 is a bit arbitrary, but there must have been a reason why Knuth increased the 1 to 10.

@mickep Maybe issues with \raggedright cases where you don't need to fill the lines?

@yo' Maybe. I guess it is described somewhere. I did not find anything in Digital typography...

@mickep actually for very large values (much larger than the badness of a line) it seems that the sign doesn't really matter. the contributed demerits are (l+b)^2 so in a typical line where b is small it's essentially l^2 so you get more demerits for having more lines but the sign doesn't matter much...

@DavidCarlisle That is true. I don't know if negative l was ever considered by Knuth.

Maybe linepenalty is a bit misleading name. "basebadness" or "extrabadness" or something could have been more natural.

@egreg for once, I didn't mind you getting the tick, I don't really want +15 at the moment :-)

@DavidCarlisle You could have gone for two accepted answers. Don't be so generous.

@DavidCarlisle 290 to go!

@DavidCarlisle hmm. I'm breaking it somewhere else, then.

@JosephWright ^^ 18 million

@yo' Congrats, even though that's not a palindrome :)

@JosephWright Did you see the siunitx question over at topanswers? I'm wondering if there is a cleaner solution than my dirty hack...

@samcarter We pass through palindromes way too quickly :)

@yo' You could give every user, who's signing up creates a palindrome, a duck as avatar :P

@samcarter haha! That might not be technically possible. Very likely the db count changes before your query result on the count is delivered.

@mickep @DavidCarlisle makes up for it by being mean to ducks.

if you have, say, two definitions of a macro, do you use two macro environments? or am I meant to only do that once?

@cfr All dodos I know are on the ducks side.

@mickep and they are all against the ducks, too.

@samcarter I know I'm not Joseph, but apart from you not using the document level interface (which I'd do in this case) your solution looks fine. siunitx uses \text{μ} (basically, in fact it uses \codepoint_generate:nn) by default in LuaLaTeX which is why the Arial μ is picked up. Since the user wants the maths font you need to define it in such a way which your solution does straightforwardly.

@cfr Well, now the glass became half empty... :)

@cfr I think either, put them in the same one if defined together but I don't think anything too bad happens if you have the name repeated although it might upset the index a bit, I don't think I ever tried

@DavidCarlisle it works. I just wondered if there was a convention because I obviously end up with two underlined references in the index for a single macro. obviously that's what you'd expect. I just didn't know if that was the way it was meant to be done.

@mickep sorry :(. on the positive side, I hear both Pegasus and Super Pegasus are for the ducks.