While updating some Fossil docs, I thought I'd mention drh's own lightweight web server, but I can't remember what it's called or where it's hosted. I tried web searches, recursive greps on the Fossil and SQLite source trees, and spelunking in drh's home pages. I even tried looking at HTTP headers in my browser's development tools looking for it to announce its name.

What is it called, and is it in fact proxying for Fossil these days on the public Fossil and SQLite web sites? I thought it was called out in the www/server.wiki doc at some point, but I don't see it.

(This is in support of footnote 2 on the Fossil vs Git doc which I've put quite a lot of polish into over the past several hours.)

It's althttpd. It's mentioned on the SSL page.

Thank you! That's two things I forgot: not only the name of the thing, but also about the docsrc repo, which I didn't even search.

Can I take from that repo name that althttpd is mainly used on the SQLite and Fossil sites to host the built SQLite docs? Does it have any function under fossil-scm.org?

Again, I'm trying to make that footnote accurate, so another way to ask is, are there any parts to fossil-scm.org which are not artifacts in the main repo or this forum repo, or which are unversioned content in one of those two repos? Is Fossil naked-to-the-world on fossil-scm.org?

drh will probably need to reply for any accuracy, but there is the selfhost page that talks about how Fossil self hosting was configured as of that document date.

Does it have any function under fossil-scm.org?

Yes, I think the path is:

stunnel > althttpd > fossil. He probably uses althttpd everywhere, but that's just a guess.

Is Fossil naked-to-the-world on fossil-scm.org?

Definitely not, the site is serving https. My understanding comes from server.wiki:

Fossil's built-in HTTP server (e.g. "fossil server") does not support TLS

Improved documentation for althttpd: https://sqlite.org/docsrc/doc/trunk/misc/althttpd.md

Is Fossil naked-to-the-world on fossil-scm.org?

Definitely not, the site is serving https.

Yeah, I get that. What I mean is, are there static resources on fossil-scm.org that are served entirely by althttpd or does every HTTP hit on that site involve Fossil at some point eventually?

Put another way, if you clone both the /fossil and /forum repos with -u so you also get all of the unversioned content in each repo, is there anything hosted on fossil-scm.org you are not getting, other than the bits of each repo that Fossil treats as private?

I'm hoping the answer is "no" so that in principle one could stand up a replacement for fossil-scm.org simply from clones of those two repos. If there are static resources served directly via althttpd on fossil-scm.org, those should probably be documented somewhere so a person could make a complete mirror, not missing anything.

I'm certain the answer is not "no" on sqlite.org, if only because of the SQLite docs. My questions are not asking about that, though. I'm concerned only with Fossil's self-hosting repo and its forum repo here.

There is some legacy static content on that site that you can reach if you know the right URL. But (as far as I know) you cannot reach any static content by following links from the homepage. Everything should be going through Fossil at this point.

Oh - there is also the content in http://fossil-scm.org/.well-known/... that letsencrypt.org uses to verify control of the website. That content is required to renew the TSL cert.

The backup sites at https://www2.fossil-scm.org/ and https://www3.fossil-scm.org/ are legacy-free and on those sites everything goes through Fossil, except for the .well-known files used by letsencrypt.org.

That's what I was hoping. The footnote's text can stand as-is, then.

If there was substantial static content, I'd have felt the need to mention it there. It's a good feeling for a project to be able to say, "Clone this and you've got everything." It's good for the old bus factor, if nothing else, particularly with small single-developer operations like my own public repos.

Since it’s not in the document, and since I’m helplessly curious: what were the main priorities that led to writing a new web server, rather than using an existing one like apache or nginx?

That's just how Richard rolls. He also programs using an editor he programmed himself (unfortunately, i'm not finding the link to it).