7

Can please someone explain to me the diffrent behavior of this code

(6 - 3);
(5 - 3);
(4 - 3);
{%%%, %%, %}

Out= {3, 2, 1}

and this one

a;
b;
c;
(6 - 3);(*comment*)
(5 - 3);(*comment*)
(4 - 3);(*comment*)
{%%%, %%, %}

Out= {a, b, c}

? I thought it should be the same. Do the comments block the outputs to be used with %?

EDIT:

I found a working case ... even if one adds spaces between the comments.

a;
b;
c;
(6 - 3);
(* comment *)
(5 - 3);
(* comment *)
(4 - 3);
(* comment *)
{%%%, %%, %}

Out= {3, 2, 1}
Phab
  • 1,623
  • 9
  • 15
  • Run the second case with a new kernel. The output is {Out[0],3,2}. The {3,2,1} in the second case is the output of the first case. Not sure what is happening but your two cases are colliding. – Ymareth Jan 14 '14 at 10:31
  • @Ymareth Right, but what's about the comments? Without them everything runs fine. Do the comments block the output to be used with %? – Phab Jan 14 '14 at 10:36
  • 3
    You have some whacky invisible characters in there somehow: when I c&p the second example, it pastes as one line. When I type in manually, works as expected. Whatever the case, it's making MM think the stuff is on one line, however it's displaying, so the %% is the prior whole result, hence the 'stacking' of the results together. – ciao Jan 14 '14 at 10:46
  • @rasher you're right, but if I type a space between the semicolon and the bracket of the comment, it has the old behaviour. And even if I delete the space again, it stucks to his behaviour. How should one edit his comments with this happening? – Phab Jan 14 '14 at 10:53
  • @Phab: I've no idea how the post contents got whacked if that's the case here, but regardless, when I type it in manually to my MM it works as expected, and as you'd expect. Perhaps start a fresh MM session, and just type it in again, see if it works like you correctly expect it to? – ciao Jan 14 '14 at 10:58
  • @rasher I added a new case. If I put every comment in a new line it works, even if I add spaces. ... so in future I better put my comments in a new line. – Phab Jan 14 '14 at 11:01

2 Answers2

9

If you examine the Cell expressions (select the cell and hit ctrl-shift-E) you can see that in the first case the lines are separated by \n whereas in the second case [IndentingNewLine] sneaks in (I pasted the second case and hit return between the lines as it pasted as a single line). It would appear that [IndentingNewLine] doesn't count as an input separator so when the evaluation runs it treats all 3 lines as one, that line ends in a ; so no output is produced.

Manually editing the cell expression to use \n instead produces the expected evaluation.

When I simply paste the second case, the code looks like 

a;
b;
c;
(6 - 3);(*comment*)(5 - 3);(*comment*)(4 - 3);(*comment*)
{%%%, %%, %}

but the only lines with a \n at the end are the first 3. I therefore conclude (speculate) that missing \ns are the issue here. Not sure why they occur (or not) though.

Ymareth
  • 4,741
  • 20
  • 28
  • +1 for digging under the covers on this. I suspect a bug in auto-formatting, though turning off a bunch of it had no effect but I didn't exhaust options. OP should file a (possible) bug report w/ WRI. – ciao Jan 14 '14 at 13:13
  • there s no "\n" in my file here, I just get "[IndentingNewLine]". But I have empty lines between. If I manualy delete them, I'll get the expected result. – Phab Jan 14 '14 at 13:22
3

Comments do block the outputs when using %, but to avoid the weird effect you found adding space after the CompoundExpression (;) I suggest you to put the semicolon after the comment. In this way the comment block % independently of blank spaces.

a;
b;
c;
(6 - 3) (*comment*);
(5 - 3) (*comment*);
(4 - 3) (*comment*);
{Out[-3], Out[-2], Out[-1]}

{a,b,c}
sam84
  • 497
  • 6
  • 15
  • 1
    Well done! That's an answer I like! But the explanation still doesn't satisfy me. – Phab Jan 14 '14 at 12:12
  • @Phab In fact, I do not understand why spaces and lines affect the % command. But I suspect that it is some type of bug, and in this case there is a satisfying explanation :-) – sam84 Jan 14 '14 at 12:43
  • So, note for future notebooks: comments in an extra line, or CompoundExpression(;) after the comment. – Phab Jan 14 '14 at 12:45
  • 1
    @Phab A better idea would be to avoid using %%%, %%, etc. A notebook is not necessarily sequential... evaluating a cell at the end of the notebook and then evaluating % at the top will return the output from the last cell. If you revisit your notebook a few months later, you won't remember what that % refers to. I'm not saying that you should avoid it completely... it can be handy, but it would be better (in the long run) if you didn't rely on it. :) – rm -rf Jan 14 '14 at 15:42