To emphasise the benefit of :issue:, I've found a few cases where the number in the anchor text and the number in the link don't match.

The code's in the repo. All 105 lines of it.

I mean, we could rewrite those macros, but why?

I mean, we could rewrite those macros, but why?

Fair enough.

I have an alergy to sphinx extensions / macros because:

1. I don't like having non-standard rst
2. Any code we put in will break at some point (got bitten by this a few
   days ago: [MRG+1] DOC: fix the copybutton on the code blocks #7634)

:) But these:

1. Are fairly simple implementations
2. make the RST much more readable: it's hard to even match parentheses visually with the former version
3. reduce the rate of error for entering issue links substantially (I found many with numbers mismatched, with _ missing from the end, with # missing from the anchor text, etc.)
4. reduce the effort for entering issue links substantially, which is really the annoyance I wanted to save when adding what's new after the fact.
5. are really easy to grep for to check what's new coverage of issues in a release

Honestly, no one checks what's new rendering in review, so we should at least make it easy syntactically to add entries!

As a concession to your very valid remarks, @GaelVaroquaux, I have crossed out the possibility of using the :user: macro. We can even delete it from the code if you insist!

test-sphinxext had been failing due to a missing docutils dependency. Have worked around

The readability is indeed much better, yet I have no opinion upon adding rst macro.

yes please, I just fixed many of them manually.

I would also be +1 on the user role because a) the list at the bottom will get pretty long and b) people don't know how the references work and add a link to a name that is not declared at the bottom.

I would also be +1 on the user role because a) the list at the bottom will get pretty long and b) people don't know how the references work and add a link to a name that is not declared at the bottom.

Shh, Andy, one change at a time and we'll get this through the censors :P

Anyone have an easy way to fix up the line wrapping, or do I need to think?

I am much more worried by the core codebase getting too complex than by
this.

Yes, I appreciate this. Some of the complexity is part of maturation. Some of it is having accepted too much breadth that we're beginning to struggle to maintain.

I've neatened up wrapping etc. Merge or rejection welcome before rebase gets out of hand.

+1

I had two PRs within the last two days that had missing links for users...

@ogrisel @agramfort @raghavrv wdyt?

But the extension doesn't currently support user links with custom anchor
text, so this PR isn't ready to do that yet.

On 21 October 2016 at 01:13, Andreas Mueller notifications@github.com
wrote:

@ogrisel https://github.com/ogrisel @agramfort
https://github.com/agramfort @raghavrv https://github.com/raghavrv
wdyt?

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#7657 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAEz641JKSEZxFWJ-PBuXIBuQwzUyc1Xks5q13b5gaJpZM4KVUEr
.

I am +1 for this because

• when one adds an issue information to a whatsnew entry one tends to copy that issue linking format and forgets to update either the link text or the link address (most often the address). This will help avoid such inconsistency.
• Plus it's easier on the eyes.

@jnothman right. I'll send a patch upstream?

I'm happy to merge this as-is though

Let's do the merge and maybe make an issue for that later?

@raghavrv @jnothman made an issue here: sloria/sphinx-issues#1 and I made a PR here: sloria/sphinx-issues#3

I am 👍 for the merge now. This makes the what's new page much more readable.

Merging. @amueller, I think it would be a good idea to backport changes into 0.18.X

yeah, but I feel like we should update the version with the one that is currently in upstream. This one doesn't include my patch for the user names, right?

But you made no changes to the feature used here. I've made an issue for #7748 :author: use. I don't see the point in needing more rebases along the way.