doc: update instructions for installing from source #5755
Conversation
xxchan
force-pushed
the
xxchan/living-moose
branch
from
bd2d5ab to
3ba2529
Compare
|
Thank you for the PR! I'll think on this some, but suspect it's unlikely we'll merge. Installation is unnecessary (and discouraged) for rustfmt contributors, and strongly discouraged for users for many reasons. There's multiple reasons for this, not the least of which being that we sometimes break up larger changes into smaller PRs for easier review/discussion and using the interim, unreleased versions can fail to adhere to the formatting stability guarantee |
|
I can’t agree with you. Installing can let a contributor or user test a PR
on their larger codebase. How could it be done without installing? I didn’t
find any other contribution doc talking about it. Besides, rustfmt iterates
quite slowly now so it’s the only way to try a new feature without waiting
for months.
Anyway, if you strongly don’t want to users to install, can you remove the
installation instructions and add some notice? I think my PR is at least
much better than that :)
…On Mon, 19 Jun 2023 at 02:09, Caleb Cartwright ***@***.***> wrote:
Thank you for the PR! I'll think on this some, but suspect it's unlikely
we'll merge.
Installation is unnecessary (and discouraged) for rustfmt contributors,
and strongly discouraged for users for many reasons. There's multiple
reasons for this, not the least of which being that we sometimes break up
larger changes into smaller PRs for easier review/discussion and using the
interim, unreleased versions can fail to adhere to the formatting stability
guarantee
—
Reply to this email directly, view it on GitHub
<#5755 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AJBQZNKQHTABSWRC57FDGG3XL6KFDANCNFSM6AAAAAAXJN7WZA>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Again, thanks for taking the time to open a PR and sharing your perspective. However, I'm not keen on getting into a debate over this. I am similarly not going to take your bait around the current rustfmt volunteers being too slow for your liking, and not volunteering enough of our free time. As the maintainers of the project, I think it's more than reasonable for us to have the final say on how we encourage, and discourage, users and potential contributors on something as foundational as installation. Running a version of rustfmt from source code as part of a test or hacking on the rustfmt codebase is entirely possible and actually pretty easy. It's something those of us contributing regularly do quite commonly as part of our inner dev loop, and something that's also done on each and every CI run. An example of one (of several) options using a basic cargo flow that I'd imagine most Rust devs are familiar with if they've ever worked with any type of bin target: Yes, I recognize that your proposed steps are useful instructions if someone is actually trying to install a version of rustfmt from source, but again my point is that it's our perspective that folks shouldn't be doing that regardless and we don't want to encourage it. Overhauling the docs is one of the myriad things on my todo list, and part of that will include the removal, or at least removing the emphasis/prominence of the from-source installation steps. If that's something you'd be willing to submit a patch for then we can get that merged (and would be most appreciative!). However, I'm going to close this now as I don't want to move in this direction. |
|
Thanks for your reply. I’m not aware of the --manifest-path option and
accustomed to running cargo fmt without args in a workspace. I can add some
docs for that.
…On Mon, 19 Jun 2023 at 15:05, Caleb Cartwright ***@***.***> wrote:
Again, thanks for taking the time to open a PR and sharing your
perspective. However, I'm not keen on getting into a debate over this. I am
similarly not going to take your bait around the current rustfmt volunteers
being too slow for your liking, and not volunteering enough of our free
time.
As the maintainers of the project, I think it's more than reasonable for
us to have the final say on how we encourage, and discourage, users and
potential contributors on something as foundational as installation.
Running a version of rustfmt from source code as part of a test or hacking
on the rustfmt codebase is entirely possible and actually pretty easy. It's
something those of us contributing regularly do quite commonly as part of
our inner dev loop, and something that's also done on each and every CI run.
An example of one (of several) options using a basic cargo flow that I'd
imagine most Rust devs are familiar with if they've ever worked with any
type of bin target: cargo run --bin cargo-fmt -- --check --manifest-path
path/to/project/you/want2test/Cargo.toml
Yes, I recognize that your proposed steps are useful instructions if
someone *is* actually trying to install a version of rustfmt from source,
but again my point is that it's our perspective that folks shouldn't be
doing that regardless and we don't want to encourage it.
Overhauling the docs is one of the myriad things on my todo list, and part
of that will include the removal, or at least removing the
emphasis/prominence of the from-source installation steps. If that's
something you'd be willing to submit a patch for then we can get that
merged (and would be most appreciative!). However, I'm going to close this
now as I don't want to move in this direction.
—
Reply to this email directly, view it on GitHub
<#5755 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AJBQZNML34FB36UP5FKAEADXMBFBRANCNFSM6AAAAAAXJN7WZA>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
|
Sure thing!
Yeah |
Copied from Clippy doc