RFC: abstract struct and abstract enum

[glaebhoerl]

No description provided.

[jfager]

'opaque'?

[glaebhoerl]

is a word I did find but didn't think is better. Perhaps other people have different opinions?

[chris-morgan]

Just a note on formatting—you don't appear to be fond of second-level headings?

Going from h1 to h3 doesn't make sense semantically.

[glaebhoerl]

@chris-morgan The semantic content of document markup is not something I've been thinking about. I couldn't tell you why I did it this way because I literally didn't have a single conscious thought about it. For some reason I felt like it. If I want to rationalize, then maybe I wanted to better emphasize the separation between the levels (I also went from h3 to h5). In any case, noted for the future.

(Aside: going from h1 to h3 to h5 is perfectly sensible if you're a rook.)

[alexcrichton]

As a side note, can you format the document to an 80-character width?

[alexcrichton]

Disallowing fine-grained field privacy seems quite unfortunate. While rare, it has some crucial use-cases (such as the fmt::Formatter struct).

[glaebhoerl]

This isn't an expressiveness question, just a convenience question. You can still accomplish the same thing by making a separate abstract struct with the private fields and including it as a field of the non-abstract struct. As I commented elsewhere, I'm also perfectly fine with adding back pub fields if people think it's used frequently enough to be worth having.

[glaebhoerl]

OK, will do.

[huonw]

(If we were to do sentence-per-line, or at least, new sentences start a new line; then the diffs might be nicer as we edit as they won't include epic rewrapping due to a 2 word addition at the start of a paragraph?)

[glaebhoerl]

I'm more concerned by the fact that after I pushed the updated version, @alexcrichton's earlier comments got hidden as "commented on an outdated diff". Given that the most important part of these RFC PRs is supposed to be the discussion, this seems unfortunate.

[nikomatsakis]

On Tue, Mar 18, 2014 at 01:45:18AM -0700, Gábor Lehel wrote:

I'm more concerned by the fact that after I pushed the updated version, @alexcrichton's earlier comments got hidden as "commented on an outdated diff". Given that the most important part of these RFC PRs is supposed to be the discussion, this seems unfortunate.

I think best practice is not to force push RFCs (that is, do not use
commit --amend, do not rebase, etc). We should probably mention that
somewhere.

[pnkfelix]

Filed as issue #14 against RFC 0001.

[glaebhoerl]

@nikomatsakis I deliberately avoided touching the existing commit (force pushing), and instead added a new one. The issue is that even so, GitHub now hides diff comments on the previous commit by default. They are still there and can be expanded by clicking on them, but they're collapsed by default, even though their content is no less relevant than it was.

[pnkfelix]

There are bookmarklets one can use to dynamically expand collapsed comments. See e.g. https://coderwall.com/p/akdgoq

(But I guess you might have to install such a bookmarklet locally on your own browser, since it seems like github's markdown renderer strips out javascript: links. Also, the bookmarklet linked there does not seem to be working out-of-the-box for me.)

[nikomatsakis]

On Tue, Mar 18, 2014 at 04:26:04AM -0700, Gábor Lehel wrote:

@nikomatsakis I deliberately avoided touching the existing commit (force pushing), and instead added a new one. The issue is that even so, GitHub now hides diff comments on the previous commit by default. They are still there and can be expanded by clicking on them, but they're collapsed by default, even though their content is no less relevant than it was.

Oh. That is certainly something to be aware of.

[brson]

This was discussed along with #1 in today's meeting: https://github.com/rust-lang/meeting-minutes/blob/master/weekly-meetings/2014-03-25.md#private-fields-by-default

The consensus was that this approach has a broader and less understood impact than simply flipping the defaults per #1; we'll continue with that approach and not this one.

Thank you.