Codeberg experiment invitation

[Wessel, Keith]

All,

I dug into this a little more with some sighted assistance. The file browser was, in fact, right there in the table that I mentioned in my email yesterday. What originally seemed like a table of recent commits was, in fact, the table with the file browser. Where I got confused was that the top row of the table doesn't contain the first file or directory, but rather information about the commit associated with the point in time of the files I'm currently browsing. You'll see that in Codeberg on the line above the first folder. This also means that the table has no header row. If it had, it might have been more obvious that I was looking at a list of files and folders in that table. I tried navigating the table yesterday but, for a reason I can't explain, totally missed the list of files and directories.

While the lack of a header row is unfortunate and would be a nice addition for Codeberg to make to their site, it's no worst than the hosted Git web interface. I think it was just more noticeable because the rest of their interface is so much better than Git web. With Git web, by the time I got to that point on the page, it was already so terrible that I was overcompensating to find what I needed.

Codeberg has some room for improvement. I'd give it a solid B+ for accessibility. That's up from the C+ I'd give the Git web interface.

But I'm going to say we're good enough here, and given the opportunity, I'd be happy to make suggestions to Codeberg to make it even better.

Keith


-----Original Message-----
From: Ian Young <ian at iay.org.uk> 
Sent: Friday, September 26, 2025 6:40 AM
To: Shib Dev <dev at shibboleth.net>
Cc: Wessel, Keith <kwessel at illinois.edu>
Subject: Re: Codeberg experiment invitation



> On 25 Sep 2025, at 21:54, Wessel, Keith via dev <dev at shibboleth.net> wrote:
> 
> All, for what it's worth, and considering I might be the only one who this impacts,

Thanks for the response. Observations that affect a smaller number of people are very relevant to a review like this.


> the web interface for Codebook is much less cluttered and easier to interact with using a screen reader than the existing hosted Git web interface.

That’s good to know. Getting to the point where we might be able to dump GitWeb is high on my list of desirable outcomes here.


> With that said, I had no problem browsing commits, branches, and tags and even searching for a file. But the one feature I couldn't do with a screen reader in a few minutes of exploration was to browse the file tree of a branch. That's a fairly basic and essential task, and while I usually just check out the repo so I can browse, search and tinker from a much more comfortable command line interface, I have been known to use the file browser in Git web. For Codebook, even though I was on the code tab, I couldn't find a list of files or a tree to traverse. I'm hoping I'm just missing something obvious rather than it being there but totally invisible to the screen reader.

I’m not sure whether you’re saying that the file tree is not accessible in general, or if that’s just true for a branch other than “main”.

If you’re at this URL, for example, can you navigate the file tree?

https://codeberg.org/Shibboleth/java-metadata-aggregator


From there, I can go through to branch selection and get to here:

https://codeberg.org/Shibboleth/java-metadata-aggregator/src/branch/maint-0.10

For me, that seems to have an identical looking set of files and if the first is accessible I’d have thought that would be too. But the underlying issue isn’t something I’m experiencing so I may be misunderstanding your point.


As you may be aware, Codeberg is a deployment of a modified Forgejo, which in turn is derived from Gitea. The issue list for Forgejo is here:

https://codeberg.org/forgejo/forgejo/issues?state=open

It is quite large (1000+ open issues). There is a label for accessibility issues with 16 currently open:

https://codeberg.org/forgejo/forgejo/issues?labels=81214

I don’t see this issue there specifically although there are a couple addressing screen readers more generally. At least they recognise the need in general.  It might be worth opening an issue there and see if it’s something they recognise; browsing files does (as you imply) seem kind of basic.


> Anyway, just wanted to provide my $0.02 on the ground floor of this experiment.

Again, thanks for this.

	—- Ian