Assign PROGRAM_FILE earlier #22092
Assign PROGRAM_FILE earlier #22092
Conversation
I discovered a use case where it would be nice to be able to access `PROGRAM_FILE` while within the .juliarc.jl. The global variable is now assigned and available within `--load` file or the .juliarc.jl. Additionally, PROGRAM_FILE is now a constant.
|
Note we may not want to back port this is a breaking change if users are accessing |
|
I don't think it's good to change the behavior of juliarc this far after feature freeze, after 2 rc's. Could you describe the use case more? Making juliarc behavior depend on PROGRAM_FILE seems like a strange thing to do |
|
I needed to run |
Completely reasonable. I'll drop the backport label. |
Would this solve #21823 then? |
|
It may also be a good idea to set an ENV variable when running within the |
|
The separate PR I was referring to: #22094 |
[ci skip]
|
Added a NEWS entry since this a breaking change without a warning. Overall I think this change is still worthwhile as it makes |
|
Expanded |
test/cmdlineargs.jl
Outdated
| println(PROGRAM_FILE) | ||
| include(\"$(escape(b))\") | ||
| """ | ||
| ) |
There was a problem hiding this comment.
nitpick, but I don't think it's usually worth using up vertical space to put close parens on their own line
There was a problem hiding this comment.
I'll update this to match the Base style
test/cmdlineargs.jl
Outdated
| write(a, """ | ||
| println(@__FILE__) | ||
| println(PROGRAM_FILE) | ||
| include(\"$(escape(b))\") |
There was a problem hiding this comment.
I'll update this to use escape_string
|
Ready to squash and merge |
| write(a, """ | ||
| println(@__FILE__) | ||
| println(PROGRAM_FILE) | ||
| println(length(ARGS)) |
There was a problem hiding this comment.
we don't care about testing ARGS length any more?
There was a problem hiding this comment.
No. Originally I added this since the number of arguments could vary depending on when we would run the check. I could still include this check but it would only be for verifying that the ARGS were no longer includes the PROGRAM_FILE as ARGS[1]
There was a problem hiding this comment.
why not keep checking the new expected value?
There was a problem hiding this comment.
The tests prior to this ("test passing arguments") deals with testing the number of ARGS already. Having them in this section is redundant and just adds noise as the length(ARGS) == 0. I'll make sure that the previous section covers what these tests previously did.
There was a problem hiding this comment.
Added new tests to address this
|
Will merge later today |
|
I'll hold off on merging today since I modified the tests. Essentially, the latest changes include:
|
* Assign PROGRAM_FILE earlier I discovered a use case where it would be nice to be able to access `PROGRAM_FILE` while within the .juliarc.jl. The global variable is now assigned and available within `--load` file or the .juliarc.jl. Additionally, PROGRAM_FILE is now a constant. * Add to NEWS [ci skip] * Test startup-file and PROGRAM_FILE * Modify user home properly on Windows * Style correction and remove escape function * Expand arguments tests to include the startup-file * Refactor PROGRAM_FILE tests
I discovered a use case where it would be nice to be able to access
PROGRAM_FILEwhile within the .juliarc.jl. The global variable is now assigned and available within--loadfile or the .juliarc.jl.Additionally,
PROGRAM_FILEis now a constant.Follow up to #14114