RFC: Add support for determining name of script #14114

omus · 2015-11-24T01:57:33Z

tkelman · 2015-11-24T02:19:19Z

base/exports.jl

@@ -174,6 +174,7 @@ export
    UnicodeError,

 # Global constants and variables
+    FILE_NAME,


this should be alphabetized in each grouping

Good to know. Thanks

sjkelly · 2015-11-24T02:58:42Z

I think n+1 people have wanted this, so docs and a NEWS.md entry might be in order?

(I also wonder if this opens up module as a script functionality?)

omus · 2015-11-24T15:05:51Z

I added/updated documentation and an entry to NEWS.md.

What do people think of the name FILE_NAME to indicate the name of the script? Anyone have any better suggestions? Also, FILE_NAME is currently a non-constant global. Is this a major problem or is there a good way to solve this that I'm not thinking of?

KristofferC · 2015-11-24T15:08:24Z

Maybe use const FILE_NAME = Ref{UTF8String}()?

iamed2 · 2015-11-24T15:57:54Z

@omus:
CLI names for the script file:

Julia: programfile
Ruby: programfile
Python: file
Lua: script
Guile: file
Racket: file
Perl: programfile
Bash: script-file
Fish: help
R: file
PHP: file

Seems like PROGRAMFILE might be the best, in order to be consistent with the CLI help.

omus · 2015-11-24T17:49:13Z

I like that name better. I'll change it to PROGRAM_FILE to make it slightly more readable.

omus · 2015-11-24T19:11:50Z

After messing around with this more ARGS only has the program file removed right before the program file is run. That means that for code executing from using --load, --eval, or --print will not be able to determine the PROGRAM_FILE since it will still be included within ARGS. I was planning to initialize PROGRAM_FILE as a constant by initializing it at the last possible moment via:

global const PROGRAM_FILE = UTF8String(shift!(ARGS))

Unfortunately this won't work as I would also like to support having PROGRAM_FILE defined and usable to any external script passed into Julia.

The repercussions of this is that PROGRAM_FILE cannot be a constant and will only be assigned correctly to a filename from within the PROGRAM_FILE itself and included files.

stevengj · 2015-11-25T19:22:28Z

NEWS.md

@@ -151,6 +151,8 @@ New language features

  * `@__LINE__` special macro now available to reflect invocation source line number ([#12727]).

+  * `PROGRAM_FILE` global is now available for determining the name of the running script ([#14114]).


you need to run julia doc/NEWS-update.jl to update the cross-reference links.

Ok. Is there some Julia contribution documentation that mentions this?

if it isn't mentioned somewhere in CONTRIBUTING.md it probably should be

Hey, I just noticed that this was accidentally added to the Julia 0.4 NEWS, not the Julia 0.5 NEWS...

Fixed in 47c29a9

omus · 2015-12-03T20:44:49Z

I'll squash these commits

tkelman · 2015-12-03T21:01:41Z

was the issue with

The repercussions of this is that PROGRAM_FILE cannot be a constant and will only be assigned correctly to a filename from within the PROGRAM_FILE itself and included files.

sorted, or is it a minor concern?

omus · 2015-12-03T21:15:33Z

@tkelman I think it is a minor concern. Basically PROGRAM_FILE will be empty for the ~/.juliarc and the --load file. I could sort this problem out but that the change would involve taking the PROGRAM_FILE out of ARGS prior to running either of these file. Changing that behaviour could break things:

Here's a demonstration of the current behavior:

$ echo 'println("PROGRAM_FILE: $PROGRAM_FILE"); for (i, arg) in enumerate(ARGS); println("$i: $arg"); end' > script.jl

$ julia script.jl foo -bar --baz
PROGRAM_FILE: script.jl
1: foo
2: -bar
3: --baz

$ julia -L script.jl script.jl foo -bar --baz
PROGRAM_FILE: 
1: script.jl
2: foo
3: -bar
4: --baz
PROGRAM_FILE: script.jl
1: foo
2: -bar
3: --baz

What do you think?

tkelman · 2015-12-03T21:34:07Z

I guess it not being const means people can modify it which could be confusing or abused?

tkelman · 2015-12-03T21:36:14Z

doc/stdlib/constants.rst

@@ -14,6 +14,10 @@ Constants
   A symbol representing the name of the operating system. Possible values
   are ``:Linux``, ``:Darwin`` (OS X), or ``:Windows``.

+.. data:: PROGRAM_FILE
+
+   A string containing the script name passed to Julia.


It would be good to clarify here exactly how this differs from @__FILE__

Good point. I'll add some more details.

omus · 2015-12-03T21:41:21Z

Yes it could be modified. I'll note that ARGS elements can also be modified if someone wants to. Alternatively I could make PROGRAM_FILE constant without changing anything else. This would mean that if someone tried using PROGRAM_FILE in the --load file it would raise an UndefVarError.

omus · 2015-12-03T22:07:25Z

@tkelman I updated the documentation to be clearer. I noticed however that macro's don't seem to support permalinks which means the @__FILE__ link I included in the PROGRAM_FILE documentation doesn't work.

tkelman · 2015-12-04T00:13:04Z

doc/stdlib/file.rst

@@ -320,7 +320,7 @@

   .. Docstring generated from Julia source

-   ``@__FILE__`` expands to a string with the absolute path and file name of the script being run. Returns ``nothing`` if run from a REPL or an empty string if evaluated by ``julia -e <expr>``\ .
+   ``@__FILE__`` expands to a string with the absolute path and file name of the containing the macro. Returns ``nothing`` if run from a REPL or an empty string if evaluated by ``julia -e <expr>``\ . Alternatively see :data:`PROGRAM_FILE`.


this is autogenerated so should be changed at the jl docstring source

omus · 2015-12-04T16:31:08Z

I tracked down the permalink issue with @__FILE__. Will be fixed when JuliaLang/JuliaDoc#18 is merged. No changes will be required in this PR.

omus · 2015-12-16T16:16:05Z

Rebased on to latest master. I also separated the PROGRAM_FILE tests from the ARGS tests.

JeffBezanson · 2015-12-29T22:45:03Z

LGTM. @StefanKarpinski what do you think?

stevengj · 2015-12-30T22:48:13Z

LGTM.

Note that PROGRAM_FILE was chosen as the name for this variable as it matches the help for the Julia CLI.

omus · 2016-01-27T17:57:02Z

Rebased again on to the latest master. Fixed conflict with 7d5def6

omus · 2016-01-27T20:01:36Z

The appveyor 64-bit Windows failure looks like it just got killed at 1 hour.

RFC: Add support for determining name of script

omus · 2016-01-28T17:00:13Z

Thanks!

…ther than 0.4) to NEWS

tkelman reviewed Nov 24, 2015
View reviewed changes

stevengj reviewed Nov 25, 2015
View reviewed changes

omus mentioned this pull request Nov 26, 2015

Add documentation on updating references in NEWS.md #14162

Merged

omus force-pushed the filename branch from 8d86da8 to 9315f9b Compare December 3, 2015 20:49

tkelman reviewed Dec 3, 2015
View reviewed changes

omus force-pushed the filename branch from 9315f9b to af0b73e Compare December 3, 2015 22:06

omus changed the title ~~Add support for determining name of script~~ RFC: Add support for determining name of script Dec 3, 2015

omus force-pushed the filename branch from af0b73e to 3fd548d Compare December 3, 2015 23:44

tkelman reviewed Dec 4, 2015
View reviewed changes

omus force-pushed the filename branch from 3fd548d to 406c6a1 Compare December 4, 2015 02:23

omus force-pushed the filename branch from 406c6a1 to 9ac75c6 Compare December 16, 2015 16:13

omus force-pushed the filename branch from 9ac75c6 to 1e75338 Compare December 16, 2015 19:51

omus force-pushed the filename branch from 1e75338 to 125c9b7 Compare January 27, 2016 17:50

Add support for determining name of script JuliaLang#14109

ffb3458

Note that PROGRAM_FILE was chosen as the name for this variable as it matches the help for the Julia CLI.

omus force-pushed the filename branch from 125c9b7 to ffb3458 Compare January 27, 2016 17:56

stevengj added a commit that referenced this pull request Jan 28, 2016

Merge pull request #14114 from omus/filename

b210456

RFC: Add support for determining name of script

stevengj merged commit b210456 into JuliaLang:master Jan 28, 2016

omus deleted the filename branch January 28, 2016 17:00

stevengj added a commit that referenced this pull request Apr 8, 2016

moved #14114 from HISTORY file (it was accidentally listed for 0.4 ra…

47c29a9

…ther than 0.4) to NEWS

omus mentioned this pull request May 26, 2017

Assign PROGRAM_FILE earlier #22092

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

RFC: Add support for determining name of script #14114

RFC: Add support for determining name of script #14114

omus commented Nov 24, 2015

tkelman Nov 24, 2015

omus Nov 24, 2015

sjkelly commented Nov 24, 2015

omus commented Nov 24, 2015

KristofferC commented Nov 24, 2015

iamed2 commented Nov 24, 2015

omus commented Nov 24, 2015

omus commented Nov 24, 2015

stevengj Nov 25, 2015

omus Nov 25, 2015

tkelman Nov 25, 2015

stevengj Apr 8, 2016

stevengj Apr 8, 2016

omus commented Dec 3, 2015

tkelman commented Dec 3, 2015

omus commented Dec 3, 2015

tkelman commented Dec 3, 2015

tkelman Dec 3, 2015

omus Dec 3, 2015

omus commented Dec 3, 2015

omus commented Dec 3, 2015

tkelman Dec 4, 2015

omus commented Dec 4, 2015

omus commented Dec 16, 2015

JeffBezanson commented Dec 29, 2015

stevengj commented Dec 30, 2015

omus commented Jan 27, 2016

omus commented Jan 27, 2016

omus commented Jan 28, 2016

		@@ -151,6 +151,8 @@ New language features

		* `@__LINE__` special macro now available to reflect invocation source line number ([#12727]).

		* `PROGRAM_FILE` global is now available for determining the name of the running script ([#14114]).

RFC: Add support for determining name of script #14114

RFC: Add support for determining name of script #14114

Conversation

omus commented Nov 24, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sjkelly commented Nov 24, 2015

omus commented Nov 24, 2015

KristofferC commented Nov 24, 2015

iamed2 commented Nov 24, 2015

omus commented Nov 24, 2015

omus commented Nov 24, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

omus commented Dec 3, 2015

tkelman commented Dec 3, 2015

omus commented Dec 3, 2015

tkelman commented Dec 3, 2015

Choose a reason for hiding this comment

Choose a reason for hiding this comment

omus commented Dec 3, 2015

omus commented Dec 3, 2015

Choose a reason for hiding this comment

omus commented Dec 4, 2015

omus commented Dec 16, 2015

JeffBezanson commented Dec 29, 2015

stevengj commented Dec 30, 2015

omus commented Jan 27, 2016

omus commented Jan 27, 2016

omus commented Jan 28, 2016