RFC: overhaul of "Scope" manual section

[mauro3]

Trying to fix issues #9469 and #9955, this turned into quite a substantial re-write of the "Scope" section of the manual. I hope this is ok, please give feedback. I do think that documenting global scope is addressed (#9469), as is the difference between soft and hard local scope (#9955), and the whole section maybe a bit more logically structured now. A semi rendered version is here.

Question: is the statement about global never having any effect in soft local scopes correct?

I'm not 100% happy with a few things but I ran out of time & steam (and expertise) to address them. If you got quick fixes then let me know, otherwise they might be left for another day. "Issues" left:

• description of let could use some love. In particular it is not quite clear what the difference is between introducing variables in its head vs. body.
• macros could probably do with an example and a link to the somewhat related topic of hygiene
• section on "Constants" seems a bit out of place. But where to move it?
• running the doctests I got lots of errors from everywhere. Is that expected because on my path is Julia-0.3? Or are they indeed broken?
• some description/link for scope in parallel programs (edit: added)

Cc'ing people from the issues and some of the original authors: @elextr , @uvtc, @JeffBezanson, @Sisyphuss, @nolta

(edited)

[mauro3]

Some updates after a julia-user conversation: https://groups.google.com/d/msg/julia-users/R9eVLw1PBj0/87ZrVib_dhoJ

[mauro3]

Now that most of the 0.4 frenzy is over, maybe this can be reviewed now? Also, if good this should be back-ported to 0.4 (and 0.3 if that is still the done thing).

[mauro3]

Ping.

[StefanKarpinski]

@JeffBezanson, I know you're busy with the function overhaul, but you really need to review this.

[mauro3]

The function overhaul is more important but afterwards some time would be good.

[mauro3]

Ping! Maybe now would be an opportune moment to review?

[El-Zurdo-Sordo]

should this be: variable(s)? you can have a for loop with multiple iteration variables, i.e.
for i=1:1000, j=1:20

[StefanKarpinski]

Merging since this hasn't been reviewed in the 6 months since it was opened and it's more correct than what we have now.

[mauro3]

Cheers!

[simonbyrne]

the buildbots are failing: there appears to have been some trailing whitespace in there

[StefanKarpinski]

Weird. It was all green when I hit merge.

[tkelman]

dangers of ci skip, reason not to use it

[mauro3]

Yes, whitespace, sorry!

[mauro3]

PR #14899 should have it fixed + the two typos mentioned above.

[Sisyphuss]

@tkelman if ci skip is dangerous, how about av skip?

[tkelman]

If you're only touching docs, then [av skip] is safer, the whitespace check will still run on travis

[mauro3]

Should this be backported to 0.4?