Source compatibility with newer cmdliner versions - #6854
Conversation
kit-ty-kate
left a comment
There was a problem hiding this comment.
Thanks a lot! After some minor reformatting to avoid unrelated changes and simplify the diff it looks great.
Note for the remaining reviewer (@rjbou): the github "Hide whitespace" setting makes the diff straightforward
|
Thanks! I wasn't sure if this kind of thing needed a CHANGES entry -- should I just ignore the CI run for that, or do you have a recommendation for what section I should put it under ("Internal"?) |
you can ignore the CI, that's no problem, we'll do it before the last rebase/final merge. But if you want to do it before that, this one should probably go under |
45934cc to
bdfb2ee
Compare
bdfb2ee to
306a1ee
Compare
306a1ee to
ede7372
Compare
rjbou
left a comment
There was a problem hiding this comment.
It worth having operators or small function name defined for OpamCmdliner.Arg.conv* in OpamArgTools and used everywhere.
what would be the point? |
Many of these are replaced in #6868 when completions are added |
to simplify the code, the use of these function. I find it simpler to write/read an converter operator and the main variable/function call, that a plain text wrapper.
Fair enough. I prefer operators/tiny functions name for this kind of usage, but it's a personal opinion, we'll see how it goes. |
|
Ignoring windows CI failure (#7030). |
Related to issues #6424, #6849, #6853
This PR makes the necessary code changes to allow compilation under cmdliner 2.0 while keeping the existing vendored version of 1.3.0.
These changes are mostly find-and-replace type things, but they would allow (at some future point):
Arg.conv'calls this PR ads with 2.0'sArg.Conv.make, which would also allow us to specify completion handlers)To verify, change the
vendor_cmdliner.shscript torun it, and re-build.