@Szabolcs sounds ridiculous. Why don't they just provide an interface to the function that converts the string so that you can do it by yourself and get rid of the source string straight away?
@Rojo I think maybe. The documentation is remarkably non-specific about a lot of things, which is fine as long as it works the way you expect, but otherwise...
On the other hand if you talk only about completely undocumented functions then I think people have to expect that they may have bugs or not be suitable for general usage.
I have upgraded my Mathematica 9 to 10 and now I am downloading it. My computer has been reformatted and the operating system is restored. My question is "should I install the version 9 first and then run the version 10 installer? Or can I just run the version 10 installer?" Sorry it is the first time I upgraded.
Cited from Wolfram site, "OutputForm[expr] : a two‐dimensional form using only keyboard characters". My question is "what does two-dimensional form mean?"
Wonder what has happened to my v10 installation now. If I do ctrl-=, write some input and attempt to exit that input container (so it would be evaluated to an entity), kernel and FE go to busyloop.
Ehm. Now the problem went away, although it persisted over multiple Mma restarts.
Smells like nasty dependency on back-end servers.
This occurred to me also roughly 20 hours or so when I last tried to demonstrate the stuff.
@Rojo heya. I hope you don't dislike my recent answer about the *By family too much. You have a good point and I should not complain too much. I suppose there are a lot of criteria and I must say I am bit fuzzy on what is important and what is not really. Anyway, of course the most (only?) important case is the documented case and the most important criterion for the code is that it gives the correct output. Second is that it has no side effects.
Third is that the code does not give bad output if we provide Unevaluated around an argument. I feel that a bad transformation in the undocumented case should be accompanied by a message, but that is probably even less important. And then there are side effects and using Unevaluated in the undocumented case, but although this can be interesting to point out, it is probably not nice to complain.
I am hoping to update the answer in such a way that I can email support@wolfram.com and link to relevant sections for bug report. Probably I will put somewhere in there that it is not the biggest issue.
Sometimes I check how many users of mathematica.stackexchange.com there are.
I remember that a few weeks ago there were about 15 thousand and recently I've been surprised seeing that the new users are signed with numbers over 18000.
Let's check this site, the new users therein have numbers sligh...
