Web lists-archives.com

Re: [PATCH] gitsubmodules: align html and nroff lists




On Wed, May 01, 2019 at 01:32:17PM -0700, Emily Shaffer wrote:

> There appears to be a bug in the toolchain generating manpages from
> lettered lists. When a list is enumerated with letters, the resulting
> nroff shows numbers instead. Mostly this is harmless, but in the case of
> gitsubmodules, the paragraph following the list refers back to each
> bullet by letter. As a result, reading this documentation via `man
> gitsubmodules` is hard to parse - readers must infer that a bug exists
> and a refers to 1, b refers to 2, and c refers to 3 in the list above.

Yikes, I see this, too.

> The problem specifically was introduced in ad47194; previously rather
> than generating numerated lists the bulleted area was entirely
> monospaced in HTML and shown in plaintext in nroff.

I wondered briefly if ad47194 was doing the wrong thing to convert this
section, but I think its _intent_ was right. Its author just didn't
anticipate this bug. :)

> The bug seems to exist in docbook-xml - I've reported it on May 1 via
> the docbook-apps mail list - but for now it may make more sense to just
> work around the issue.

Yeah. Specifically, the HTML generated directly from asciidoc doesn't
have this bug. Likewise, using asciidoctor has no impact (though in
theory if we eventually move to generating roff directly from
asciidoctor, the problem might go away).

I agree with you that it's worth working around in the meantime.

The patch itself looks good to me. A few observations:

>  A submodule is considered active,
>  
> -  a. if `submodule.<name>.active` is set to `true`
> +  1. if `submodule.<name>.active` is set to `true`

I'd sometimes use letters when there is already a nearby list using
numbers (and I need to be able to refer to either list distinctly). But
I checked the surrounding context and don't think that's the case here.

> -Note that (c) is a historical artefact and will be ignored if the

Not introduced by your patch, but it may be worth swapping this out for
"artifact". The spelling with the "e" is apparently a British-ism (today
I learned!), but I think we usually favor (favour!?) American spellings
in our documentation.

Obviously not super important, and definitely should be addressed in a
separate patch (if at all).

-Peff