No interpolation of const values in a literal passed to `help` or `short` of `dispatch`

[ZoomRmc]

import cligen

const aH = "foo"
let bH = "bar"

proc main(aaa: string; bbb: int) =
  quit()

when isMainModule:
  dispatch(main, help = [("aaa", aH), ("bbb", bH)])

Output of nim r test.nim -r:

...
  -a=, --aaa=    string  REQUIRED  aH
  -b=, --bbb=    int     REQUIRED  bH

Expected: compile time error for using variable bH, using "foo" for aaa help. Silently inserting names of the constants/variables is even more surprising.

[c-blake]

The inserting name thing is probably just some strVal "helpfulness". I have to evaluate/translate an expression tree that the macro gets delivered and that parseHelps work is..incomplete.

[c-blake]

Oh, also - as a corollary to evaluating/translating, things that are "semantically similar in Nim" are distinct as seen by a macro. So, {a:b, c:d} delivers a different syntax tree than [(a,b),(c,d)]. I can see that this relates to your https://github.com/c-blake/cligen/issues/208 , though and I am working on some solution for you.

[ZoomRmc]

So, {a:b, c:d} delivers a different syntax tree than [(a,b),(c,d)]. I can see that this relates to your #208 , though and I am working on some solution for you.

That's interesting, considering table literal doesn't exist in the language itself. However, in this case it makes no difference, the behaviour is the same.

[c-blake]

Actually, I might be wrong about delivering a different syntax tree. It may not be possible to de-sugar that any differently. Sorry - working on all your issues at once. :-D

[c-blake]

You need to change your let to const to make your code work, but that is the other error message issue...Still working on that, but presumably it is lower priority... :-)

[ZoomRmc]

You need to change your let to const to make your code work, but that is the other error message issue...Still working on that, but presumably it is lower priority... :-)

I put "let" there to show it's a case worth having an error for. Code works as shown, but shouldn't. It should work with a var changed to a const (see the "Expected" line).

The crux of the issue was const not being treated.

[c-blake]

Well, I took the crux to be the variable name instead of the string... :-) My just committed error message code does not seem to work as well on this snippet as the other issue one...Investigating.

[c-blake]

Well, I committed a check that should work ok. Let me know what you think whenever you get a chance.

[ZoomRmc]

Works perfectly. Thanks a lot!