#dev 2024-03-01

2024-03-01 UTC
#
to2ds
Continued from #indieweb. I do have a webmention endpoint but it's full manual labor mode ATM.
#
to2ds
Can a webmention be replied to? For instance I send one out, and it exists as an explicit HTML file on my system can it be a receiver of webmentions itself?
#
aaronpk
a webmention is a notification about a post
#
aaronpk
so that post can of course receive webmentions itself
tryinbrian joined the channel
#
to2ds
Thank you. I think I was overthinking that :)
#
aaronpk
it is easy to get confused about the terminology around webmention
#
to2ds
Yes. It's very straightforward in concept but the devil is in the details.
#
[tantek]
to2ds have you tried webmention.rocks with your site/implementation?
#
to2ds
I did some manual tests of endpoint discover, but got confused on a few tests looking for a link tag, but none existed unless I missing some. Believe it was test 1 and 2.
#
to2ds
Theory is well and good but you learn the most when under duress during live testing πŸ˜„
#
gRegor
I think the HTTP Link header is optional for receiving, a <link> or <a> in the HTML is good
#
to2ds
Hope I wasn't overlooking something obvious.
#
gRegor
Like test 3 and 4 you could probably pass. Don't have to pass them in order.
#
to2ds
My endpoint discoverer thingy resolved both 3 and 4 properly.
#
gRegor
Nice!
#
to2ds
It uses BeatifulSoup and builds a collection of link and a and captures the line number to find the first one, though it's not happy when it finds two on the same line ATM.
#
gRegor
Two <link> tags?
#
to2ds
I believe it was two a tags. The test result had error at the end of the endpoint. Both were in the HTML body on the same physical line and the code picked the second one. Not sure why yet.
#
to2ds
Guessing I'm getty rusty πŸ˜„
#
[KevinMarks]
I use the mf2 parser to look for rel="webmention" in the target
#
[KevinMarks]
but mention.tech sends to all found webmention endpoints which is a bit of a spec violation
Renfield and to2ds joined the channel
#
to2ds
Completely forgot about mf2. It's in my virtualenv packages too πŸ˜‘
#
to2ds
That should have been a 🀦
burley joined the channel
#
iammnchrm
Hi everyone, complete beginner trying to set a personal website from the ground up. I've decided on the digital garden/second brain approach, since I've been already note taking in markdown for a long time, it feels like a natural transition, but every time I get I to something more technical my brain hurts. The dilemma I'm in right now is about SSG X CMS and which one to use, but I haven't been able to find beginner-friendly material
#
[tantek]
it's a valid dilemma. there really isn't a "beginner-friendly" SSG x CMS AFAIK. either involves some familiarity with code and/or command lines.
#
[tantek]
the /SSG and /CMS have some comparison tables if you want to take a look and see if perhaps there's something there that requires learning something that you don't mind spending the time to get up to speed on.
#
to2ds
Ok. Just found the bug. The routine iterates all link and a tags with rel="webmention", adds them to a dict using line number as the key. Sorts the dict by key (line number) and pulls the first one. The routine was overwriting the key if two tags with rel="webmention" were found in same line according to bs4 and ultimately picked the last one found.
#
to2ds
iammnchrm - Just knowing Markdown is a great start in and of itself. My earliest website versions used Markdown files as the source of the pages.
#
to2ds
iammnchrm - Have you ever experimented with Python?
#
iammnchrm
in college, a long time ago
#
iammnchrm
i know more than i should, but less than i need
#
to2ds
That might be sufficient to get started with a simple homegrown SSG using the Jinja2 template engine.
#
iammnchrm
not sure if it's okay to mention those by names (sorry if it is) but i've seen something about Obsidian + Quartz workflow
#
iammnchrm
i'm currently reading it's documentation
#
to2ds
Ah. It's generally best to start off simple and then go complicated πŸ˜„
#
iammnchrm
it can be pretty overwhelming for a beginner, specially because i've spent a lot of time trying to figure out things that were outdated/unecessary
#
to2ds
It's tough at first be definitely doable.
#
to2ds
be=but
#
iammnchrm
to2ds: the thing is, i kinda lack the means to define what's simple and what's complicated, everything looks complicated, (at leart for now)
#
iammnchrm
[edit] to2ds: the thing is, i kinda lack the means to define what's simple and what's complicated, everything looks complicated, (at least for now)
#
iammnchrm
that's why i came looking for a "start here, go in this direction" rather than "do this" advice
#
to2ds
That makes sense.
#
to2ds
Of course the simplest is to create HTML pages manually, but that can get tiresome eventually.
#
iammnchrm
to2ds++
#
Loqi
to2ds has 2 karma in this channel over the last year (6 in all channels)
#
iammnchrm
is this how it works?
#
iammnchrm
it is :3
gRegorLove_, lockywolf and to2ds joined the channel
#
to2ds
Not sure if my simple is the same as everyone else's simple, but for starting out with an SSG, I would want something that merges a flat file with a template to create an HTML file and emphasize content and presentation over tools per se. Something with few dependencies to minimize unnecessary distractions while learning. Once the rudiments are learned, more tech can be added as desired.
#
iammnchrm
precisely what i'm trying to do, nice to know i'm headed in the right way
#
to2ds
My first attempts used just that. One Python program of roughly 100 lines or so. Markdown files with an information section commonly known as "front matter", i.e. title, date, a meta description tag, a Jinja2 template or two for the merge, and a CSS style sheet.
#
to2ds
Sounds like you're on the right track then.
#
iammnchrm
pretty much it, YAML frontmatter on a .md file using (admittedly very simple) templates so that my private ramblings become public ramblings eventually
#
[KevinMarks]
Hugo does that well, though the templating model is a bit complex
#
to2ds
Always wanted to experiment with Hugo.
#
to2ds
iammnchrm - My CMS is built upon that simple model, only with a front end built in Flask which leverages Jinja2.
#
to2ds
The YAML front matter and Markdown were just split into database table fields eventually, but largely remained the same since the inception.
#
gRegor
what is obsidian
#
Loqi
Obsidian is a free, multi-platform application for writing and maintaining plain text notes (for longevity) as markdown files that provides a modest paid hosting service one can use with a personal-domain https://indieweb.org/Obsidian
#
gRegor
could try their publish option since you're using that already
[tw2113], river1 and to2ds joined the channel
#
to2ds
Obsidian seems to be a simpler path for starting out.
jeremycherfas, bterry, adele, mahboubine, GWG, jacky, lockywolf and [lcs] joined the channel
#
[lcs]
I've also been thinking about going down the path of a static site generator. But I've been using WordPress for almost 20 years, so I'm very attached to it.
#
[lcs]
Plus just using an existing SSG feels like exchanging WordPress for another tool. I should really build it myself, right? But then it essentially always comes down to Carl Sagan's quote about making apple pies from scratch (which requires you to first invent the universe) and so here I am.
gRegor joined the channel
#
[lcs]
While I was typing that, I started wondering: Does there exist a tool/program that you can download as an executable to your computer that allows you to generate a static website from a "local CMS" so to speak? No CLI, no coding, no technical insight required, just download, install and start making your website.
#
[lcs]
Anyone know?
#
jeremycherfas
Well, things like Tinderbox would create a static site for you, but I wouldn't describe them as "no insight required". And there are places online to which you "just" upload a file that then serve the contents, but I don't know anything in detail. Maybe there are local tools that can do that.
#
[lcs]
Thanks for the input [jeremycherfas] I was looking for the last variety. A local tool that essentially just lets you type, and then creates static (linked) files for you, that you could then just upload to any old server.
#
jeremycherfas
Yeah, sorry I can't be more help.
#
[lcs]
No worries! I feel like this should be a thing, in spirit of democratising truly owning your content.
#
jeremycherfas
I'm pretty sure it is a thing, just outside my exerience, except vaguely.
#
[lcs]
Yeah I'd be amazed if it isn't.
#
[lcs]
I suppose my comment should be that it should be a more known thing.
#
[lcs]
If it is a thing.
#
[lcs]
When you ask for an API key to use in a locally run Python script to automate some classification processes, and engineering responds with a link to https://techcommunity.microsoft.com/t5/ai-azure-ai-services-blog/azure-openai-architecture-patterns-and-implementation-steps/ba-p/3979934 and asks you to clarify if your "application" will be built according to the principles outlined in the article πŸ˜…
#
[lcs]
"Yessir, I'll most certainly ensure that my locally run Python script uses AOAI with Azure Frontdoor for loadbalancing"
#
capjamesg
πŸ˜‚
ttybitnik joined the channel
#
[Murray]
We recently had to get a QA set up on an LLM-backed project at work. One of the questions that came back from the IT team that manages our various permissions was "To the nearest 10, how many API calls will this person be making" and just... how do you even answer that? Their job is to _hammer_ the app to see where the edge cases are, so who knows, maybe a few tens of thousands... to the nearest ten πŸ˜‚
burley joined the channel
#
[lcs]
To the nearest ten πŸ˜‚ 94820!
#
[lcs]
Give or take
#
[Murray]
+/- 10 πŸ˜‰
burley, gRegor, [Jo], burley_, mahboubine, AramZS, GWG, [Ana_R], sebbu and [jamietanna] joined the channel
#
[jamietanna]
https://rss-is-dead.lol/ doesn't seem to support Bridgy Fed, so the question is whether it only works with Mastodon-like servers or there's something we need to support? (curious, not expecting changes to be made)
jacky, bterry, burley and [Joe_Crawford] joined the channel
#
sknebel
huh, apple actually reversed course on homescreen webapps
#
aaronpk
oh wow
#
sknebel
[jamietanna]: that works with aaronpks website, so its not doing anything tooo special I guess
gRegor, gRegorLove_ and to2ds joined the channel
#
to2ds
Love that bold typeface on the rss-is-dead.lol website.
[aciccarello] and [snarfed] joined the channel
#
[snarfed]
🀷
#
[snarfed]
would be funny if it was only Mastodon and a bit beyond, since Mastodon now has its own per-profile RSS feeds
#
jacky
it seems to pick those up but only if they're in the profile links
burley, gxt_, adele, jeremycherfas and [nsmsn] joined the channel
#
nsmsn
The RSS icon doesn't show up for the Fediverse profile of WordPress blogs using the ActivityPub plugin on rss-is-dead.lol
[Al_Abut], burley, [tantek] and [jeremycherfas] joined the channel
#
capjamesg
😬
#
capjamesg
I know about the safetensors format. I did not know it had been exploited.
#
capjamesg
-> #indieweb-chat
#
capjamesg
In fact, never mind.
#
capjamesg
I don't know too much about serialization, but I understand safetensors is a move in the right direction away from pickle, which the Python docs themselves explicitly say is a potential security risk. You should never open untrusted pickle files.
#
aaronpk
wow, tailscale's openid login uses a URL as the client ID too! https://login.tailscale.com/start
#
aaronpk
oh wait nevermind, i apparently did that
#
aaronpk
it is cool that they have a "bring your own IdP" in the signup flow, that is somewhat unusual!
jacky, gRegorLove_, burley and to2ds joined the channel
#
_tommys
Murray[d] I got around the container css thing with contain:inline on β€˜:host’ and the negative margin on the h1 like you mentioned - thanks for the tip!