Change author/about/version behavior
#229
Conversation
|
I decided not to create a new PR for the |
|
Interesting... |
|
@TeXitoi Could you please take a look? It seems like travis crashes on all channels with some compiler-bug error. I can't reproduce it locally, my |
|
My travis build can't reproduce it. Any idea? |
|
Strange... |
|
Yahoo! Now travis is all green just like a troll from 4chan. Can I ask you to download and share build artifacts of this particular build with me? I'd like to dig into this bug or at least report it to rustc team |
|
I can't find anything to download. Any pointer? |
structopt-derive/src/lib.rs
Outdated
| @@ -282,7 +283,7 @@ fn gen_clap(attrs: &[Attribute]) -> GenOutput { | |||
| let attrs = Attrs::from_struct(attrs, Sp::call_site(name), Sp::call_site(DEFAULT_CASING)); | |||
| let tokens = { | |||
| let name = attrs.cased_name(); | |||
| let methods = attrs.methods(); | |||
| let methods = attrs.clone().top_level_methods(); | |||
There was a problem hiding this comment.
i just realized that this is quite complicated question, let me think...
There was a problem hiding this comment.
In my previous attempt these methods mutated self. Hence we happen to reuse attrs here first call would add methods that are acceptable for clap::App but not for clap::Arg. First we used this to generate App and later we generated Arg. My current approach is different, these methods mutate self no longer.
I don't know, I can't find any way to do that in google so I asked the overmind, let's see what we can do |
Looks like an incremental bug. I think it is covered in rust-lang/rust#63161, you can leave a comment there. |
Hmm, I like to leave comments, especially on my second vacation day :-) |
|
Also, I squashed all the commits into one, tell me if you want to separate them |
|
I squash at merge time. That's simpler if I can review the modifications added, thanks. |
|
May I ask why do you always squash? |
|
One commit, one feature. And some contributors do a lot of commit as "fix typo", "fmt", "oups, bug fix" that can be annoying in a git bisect. |
|
Yeah, I do understand why people squash meaningless commits like |
|
If you do that, I'll just rebase, not squash |
|
I don't want to put pressure on contributors about managing commits in a certain way, but I might ask for a separate PR for unrelated stuff. I allowed you to do one PR for raw and change behavior, thus I will not ask you to do 2 PR, but if you separate this work in 2 commit, I'll happily |
Exactly what I would do, except I'd probably just isolate these 2 commits myself. N.B: I'm not actually arguing, I just noticed some policy that is different from mine and got curious about it's advantages and drawbacks. That's how people learn, and I'm trying. |
|
Thanks! |
|
Mm, @TeXitoi did you look at travis output? I'm asking because it seems we missed a warning, I just was about to push the fix when you merged it :) |
|
Sorry no, just looked at the status |
|
It seems like the docs are out of date with this? They seem to still indicate, in https://docs.rs/structopt/0.3.14/structopt/#magical-methods that these default to the Cargo values. Edit: Oh, no, I see it says "only when |
Closes #217