@egreg Thanks. :) Now I need to work hard for the 10k. :)
I had to go to a nearby city today and get some papers. I decided to walk, but when I was coming back, the rain started. I walked one hour in the rain. I stopped in a bar and bought a Coca-Cola can. They could call me for an ad, like walking in the heavy rain and drinking Coca-Cola. :P
@PauloCereda I wondered if it would be an easy apache rewrite for just calling texdoc.net/packagename which rewrites to show?pkg=packagename to make access easier
@PauloCereda yet I made endless recursion or no clever match
it's already nice. I made a button and bbcode for easy documentation link, [doc]packagename[/doc] becoming <a class="postlink" href="http://texdoc.net/show?pkg={SIMPLETEXT}"><img src="/texdoc/icon.png">{SIMPLETEXT}</a>
Should rounding be done as round(123.456,-2), or round(123.456,1e-2), the second standing for "round to the closest multiple of 1e-2? The second would be much more general, and seems "cleaner", but means that I need to implement rounding to a multiple of any number. In a sense, this would be a sibling of mod.
About Uiy's strange input syntax, I suspect he might be taking some input from a program that is hard/impossible to configure. Of course, the best would be to have an intermediate pass through Perl, or something.
@BrunoLeFloch Uiy's strange syntax comes from a quest to mark music code, it lacks though a proper sense of computer science (there are no proper delimiters) and would get into a big jam if there are errors and he uses regex.
@BrunoLeFloch He also needs to work on his patience and tact:)
@BrunoLeFloch On the rounding I would keep to conventions round(2.675, 2), I would also have a trim function to remove trailing zeros.
@YiannisLazarides I'd agree: usually people think in 'decimal places', with negative values representing rounding before the radix (if accepted at all): see I think numprint.
@YiannisLazarides I'd agree: I'd parse his input on a tokenwise basis, as I do for numbers in siunitx
@JosephWright Ok. That's easier for the rounding itself, but a bit messier for the parsing (the second argument will first be parsed as a floating point, then turned into an integer).
@YiannisLazarides Trimming zeros does not make sense on the floating point itself, only when outputting them. Internally, floating point numbers are simply real (well, rational) numbers, so "trailing zeros" or "leading zeros" have no meaning.
@JosephWright Yes, I mentioned LPeg somewhere :).
So I guess I should code the functions round(x,n), floor(x,n), ceil(x,n) within expressions, and \fp_to_scientific:n, \fp_to_decimal:n, \fp_to_decimal_trim_zeros:n for output?
@BrunoLeFloch Best IMHO to see what other languages like python are providing docs.python.org/library/decimal.html most newcomers to LaTeX now probably have a bit of knowledge of python, so it would be nice to have a one for one macros.,
@PauloCereda Sounds great! You could add some of the further good suggestions by other fellow contributors, making a nice overview or small digest about efficiently writing papers or small documents (books may better be differently handled)
@YiannisLazarides I think there are two aspects: rounding a floating point number in the middle of a computation (or to store the rounded value), and formatting the floating point number. The first task would be the round(123.456,1)=123.4. For the second one, I was indeed thinking of python, at the moment \fp_format:nn { 123.456 } { 3.1e }. Many of the python bells and whistles don't make sense before we get to strings, since padding with spaces when typesetting is wrong.
@YiannisLazarides Yes, hard to get! But the score can grow in some time. The harder part is to have a question with such a low score. If this question would ever drop further, I might get one ;) tex.stackexchange.com/questions/35892/…
While writing lecture notes from statistical physics I found out, that I am often writing very similar long LaTeX sequences like
\frac{\partial }{\partial <foo>}
\frac{\partial <bar>}{\partial <foo>}
and
\left( \frac{\partial <bar>}{\partial <foo>} \right)_{<...
@PauloCereda actually it was the TEXDOCS variable which included the user's home TEXMF which is /var/www/texmf for the apache user but was changed to the document root and so it became the main texmf
@PauloCereda we found it straightforward - noticing that it's only the texmf sub tree, and that it could be an apache user issue as you guessed, then investigating TeX vars, good troubleshooting
I am looking for an editor or documentation browser that can look up all commands from all packages, for eg I want to for 'whiledo' then I want a window or pane to pop up with the required reference information. Also one forgets the command so it would be helpful to type in 'loop commands' and a ...
@StefanKottwitz I read a blog post by Joseph about TeXworks dev and it seems very interesting. I just need to (re)learn C++ all over again, but it sounds like a good plan to me. :)
@egreg The original purpose was markup for forums and sites. LaTeX-Community.org now has buttons for CTAN package links, for direct documentation links (texdoc.net query), for environments and for commands (accessing the LaTeX2e reference for this)
@egreg The "portal" page is a bonus, and just a start for now
@egreg I think it's pretty good for being developed in some minutes just by the way
@egreg: one of the challenges of this idea is that it does not depend on human intervention. The site has always the bleeding edge info, as long as the underlying TeX distro is properly updated. :)
Would it be appropriate to ask a general question with a title such as "How can other programming languages be used to create content for TeX documents?" in an attempt to get a big-list of ways, e.g. answers with approaches involving LuaTeX, org-mode, make, write18 and bashful.
I do not have a specific such problem but I am interested in it.
@MarcoDaniel We have some ongoing discussion about consistency in these get functions, and one question is what to do about expandable ...get:Nn functions.
@JosephWright: I have a next stupid question. I want to understand the definition of prg_case_XXX:nnn whereby XXX can be int or dim. What does the command \tex_romannumeral:D do?
@MarcoDaniel It's the trick that causes expansion: \romannumeral starts expansion which ends when something not part of a <number> is found; the tokens that form the <number> are converted to a roman number if it is positive, otherwise the expansion is empty. It's always used for f expansion (by ensuring that a nonpositive number is eventually found).
For example, the expansion of \exp_args:Nf is \exp_after:wN #1\exp_after:wN {\tex_romannumeral:D -`0#2} and you see the negative number -`0
@JosephWright An example for inline linking to corresponding page in LaTeX2e command and environment reference manual: latex-community.org/forum/… (next step: adding links to LaTeX keywords in highlighted code in code environments)
@JosephWright Not really. :) A token just sitting after -`0 is expanded; this is a consequence of another fact which is independent of \romannumeral. :)
@MarcoDaniel -`0 is a negative number! As Joseph points out, TeX looks for a space to be ignored afterwards when it's looking for a number and expands the token immediately following any constant, even alphabetic constants such as `0
A nonexpandable token of course interrupts this process.
@egreg and @JosephWright Thanks. I have follow up question before I start programming examples. Why does the first token be expanded and not the second one? \testc isn't a constant.
@PauloCereda Great! Just finished a blog post - announcing that somebody with OS X know-how is needed for helping with technical issues so that the translation of the newest KOMA-Script manual can make progress. texblog.net/latex-archive/news/koma-script-os-x
@MarcoDaniel \romannumeral keeps expanding tokens until it finds the optional space or something non-expandable. In the example, it expands \testa to \testb\testc, then expands \testb to 123, then finds 1 which is non-expandable and so stops. \testc is therefore not expanded. There was a question a few days ago where I wrote a loop to expand using \romannumeral in a repeated sense to deal with this case.
@StefanKottwitz Do you know what the issues are? I have a Mac, but I'm not sure if I count as an OS X expert!
@MarcoDaniel Well, in \prg_case_dim:nnn the final case is to check if the number is equal to itself, which tidies up the loop and exits with the 'other' case. There, the = is hard-coded, but in your case it is not and so a relation needs to be added in.
@JosephWright I believe I see the trick the second argument of prg_relation_case_dim_aux:nnn is a list of braced groups whereby two groups belong together. The first part of the group is the relation and the second on the result. That's the reason that the #2 isn't braced. But why do we need an extra quark?
@JosephWright very funny -- cs_new:Npn \prg_case_end:nw #1 #2 \q_recursion_stop { \c_zero #1 } is documented with: The \c_zero marker stops the expansion of \romannumeral. Now The wheel has come full circle. ;-)
While writing lecture notes from statistical physics I found out, that I am often writing very similar long LaTeX sequences like
\frac{\partial }{\partial <foo>}
\frac{\partial <bar>}{\partial <foo>}
and
\left( \frac{\partial <bar>}{\partial <foo>} \right)_{<...
