#dev 2024-10-24

2024-10-24 UTC
#
[mattl]
so you can go to a profile such as https://libre.fm/user/janerationx
#
[mattl]
click on the item in the last column of the table
#
[mattl]
there's an embed link in the <head>
#
[mattl]
and the embed code contains h-embed and h-entry
#
[mattl]
303,620,131 x 2 unique new URLs added to the web, just like that.
#
Loqi
[preview] [mattl] #53 New feature! Individual scrobble pages and embeddable
barnaby joined the channel
#
[tantek]
[mattl]++ nicely done!
#
Loqi
[mattl] has 9 karma in this channel over the last year (28 in all channels)
geoffo, [Joschi_Kuphal], GuestZero, [aciccarello], [Jo], thegreekgeek_, ttybitnik, thegreekgeek, [aaronpk] and [snarfed] joined the channel
#
[mattl]
I added a ShareOpenly link this morning
barnaby, [Joe_Crawford], IWSlackGateway, [aciccarello], GuestZero, sebbu and gRegor joined the channel
#
[aciccarello]
I currently have a /sitemap.xml page. If I added a /sitemap/ page that returns HTML, is that going to mess up indexing of my site in search engines?
#
gRegor
It shouldn't
#
[aciccarello]
I'm trying to see if google automatically looks for a sitemap but I'm not seeing anything. Only manual sitemap submission.
#
gRegor
One more way for bots to crawl the pages. I've not seen anything that they'd punish for having both.
#
[aciccarello]
The html version probably wouldn't link to every single page, just the top level ones. While the XML includes everything.
#
gRegor
I tend to prefer a page like that over the .xml, unless you're trying to get a brand new site indexed initially.
#
gRegor
Ah, that makes sense
#
[aciccarello]
All these /slashes pages are inspiring me to add something for humans.
#
[aciccarello]
Also might be and excuse to simplify my footer.
#
[aciccarello]
According to this page, a robots.txt is the place to inform search engines https://www.sitemaps.org/protocol.html#informing
#
[tantek]
aciccarello you could use conneg on /sitemap just in case a search spider comes there asking for XML to return your sitemap.xml
#
[aciccarello]
Good point
#
[aciccarello]
I generally try to avoid conneg 😅
#
gRegor
conneg--
#
Loqi
conneg has -22 karma in this channel over the last year (-24 in all channels)
#
gRegor
what is sitemap
#
Loqi
A sitemap is a list of pages on a website https://indieweb.org/sitemap
#
gRegor
I've leaned on the good inter-linking instead of a sitemap.xml, since Google says the .xml is just a hint to them and doesn't guarantee being recrawled sooner
#
gRegor
Might use it if I had a new domain with a lot of pages to get in initially, otherwise probably not
#
[aciccarello]
Apparently netlify doesn't support redirects based on the accepts header
#
[aciccarello]
Only `Role`, `Language` and `Country`.
GuestZero joined the channel
#
[tantek]
[mattl] possible opportunity to advocate / offer http://libre.fm opt-in support in this WordPress plugin? https://indieweb.social/@janboddez/113363867985521055
#
[tantek]
specifically since he said "probably drop the scrobbles, too". my guess is that Scrobbles would "look better" in the IndieBlocks plugin if they supported Librefm embeds?
to2ds joined the channel
#
[mattl]
Yeah, I just asked if they're using http://Libre.fm
#
to2ds
conneg--
#
Loqi
conneg has -23 karma in this channel over the last year (-25 in all channels)
#
[mattl]
Why don't people like conneg?
#
_pi_r2_0[d]
conneg--
#
Loqi
conneg has -24 karma in this channel over the last year (-26 in all channels)
#
to2ds
AFAIK it can be problematic in certain use cases.
bterry1 joined the channel
#
_pi_r2_0[d]
[mattl] it breaks my expectation that to one URL corresponds one resource (and I don't treat 2 distinct representations of the same entity as the same resource)
#
to2ds
Not using it solved a problem for my current side project.
#
to2ds
Wasn't there an issue with caching as well?
#
_pi_r2_0[d]
I want to persist a site by mapping its paths to filesystem paths. when multiple resources exist at a given path I can't do that.
#
[tantek]
what is conneg
#
Loqi
conneg is short for HTTP Content Negotiation, a method by which a browser or other web client can request content of various types from a web server, and depending on what is requested, and what the server supports, it tries to provide the best it can https://indieweb.org/conneg
#
to2ds
I like to think of the old SNL skit "Shimmer" for a conceptual model of conneg :D
#
[mattl]
The reason I'd be curious about it would be for WebP images. I'd love to see a way to get people using next gen image formats but have a fallback for loading an image in a new tab or downloading an image.
#
[mattl]
Just found a great example for Embed
#
_pi_r2_0[d]
[mattl] you can conneg image formats, but won't be able to reorient on new tab on download because the cache(s)/server(s) can't distinguish between fetching for display embedded and fetching for full tab / download (nor is a redownload desirable) AFAICT
#
_pi_r2_0[d]
hmmm lemme double-check the behaviour of <picture><source type=…/></picture> in that regard
#
[tantek]
[mattl] have you seen the ellecordova skit about file formats (including images?)
#
_pi_r2_0[d]
Well, you can detect a tab as it'll `Accept` `text/html` first, vs an `img` or `picture > source[type=…]` or CSS `background-url` which won't. However they all seem to offer all image formats all the time in Chrome.
#
[Joe_Crawford]
On the topic of XML sitemaps I think there’s a fair amount of skepticism of their necessity but the Google console will tell you to add it via a line pointing to it in robots.txt, and if you have more than would “fit” in a single file (there’s a “maximum” of 10,000 urls per sitemap) you break it up into a sitemap that refers to other sitemaps.
#
[tantek]
sidefiles--
#
Loqi
sidefiles has -1 karma over the last year
#
[Joe_Crawford]
https://gist.github.com/artlung/613e6ac577a170bf2b4b9045486f129e is approach to that of one feels so loved
#
[Joe_Crawford]
So MOVED to do so. Thanks autocorrect.
#
[Joe_Crawford]
But 100% sidefiles--
#
Loqi
sidefiles has -2 karma over the last year
#
[Joe_Crawford]
I love the note in the sitemap page about Aaron having that file be blank and not seeing any change in Google. I turned off my sitemaps maybe 6 months ago and I’ve not noticed any better or worse ability to find things in my site in Google, DDG. Maybe too early to see an effect but I think spiders spider.
#
to2ds
Quietly checking own website for sidefiles-- :D
#
to2ds
sidefiles--
#
Loqi
sidefiles has -3 karma over the last year
#
[Joe_Crawford]
Sidefiles must prove their worth. Earn their keep.
#
[mattl]
What is sidefiles
#
Loqi
sidefiles is a term used for the practice of generating additional files or content to present a similar resource https://indieweb.org/sidefiles
#
[Joe_Crawford]
Otherwise they must feel the sting of `rm` and `git rm`
#
_pi_r2_0[d]
Love shipping a bunch of AVIFs for various display sizes and JPEGs for fallback through
#
gRegor
I learned that they support text sitemaps earlier this year and definitely thought if you need one, that should be the way. One URL per line, done.
#
_pi_r2_0[d]
does that help anything in any way vs relying on the crawler to discover everything?
#
gRegor
Not really, some earlier discussion about that in dev today
#
gRegor
Probably somewhat helpful if you're trying to get a bunch of *new* URLs in the index, or if they're not well inter-linked.
#
gRegor
Questionable otherwise, imo
#
[mattl]
XML feels very non-IndieWeb
#
to2ds
There's got to be some plusses to XML though.
#
_pi_r2_0[d]
you can put newlines in filenames
#
to2ds
There you go!
#
[snarfed]
newlines in filenames wat
#
[snarfed]
so cursed
#
to2ds
CDATA was my favorite part of XML.
#
[tantek]
if you want to maintain an illusion about XML "cleanliness", then don't look up: XML exceptions to Unicode
#
[tantek]
[mattl] XML on the web is largely relegated to: legacy Atom (feed reader) support, "podcasting". I think besides that XML is fairly uninteresting on *the web*. Web-adjacent there's eBooks
#
_pi_r2_0[d]
well, SVG?
#
to2ds
Thank goodness for Inkscape.
#
[tantek]
yep, goodcall pierre. SVG as the 3rd use-case (maybe even primary now). And maybe a fourth use-case of MathML
#
[Joe_Crawford]
Podcasts and SVG for the web, yes.
#
[Joe_Crawford]
for those of us who have programmed against apis that involved data transfer of CSV files or even the horrible fixed line length files the us government uses, XML is a step up.
#
[tantek]
RSS is not really XML in practice. In practice RSS consuming code must implement their own RSS tag soup parsing heuristics. If you attempt coding consuming RSS as XML, you're going to have a bad day with most of the RSS out there. [KevinMarks] can confirm.
#
gRegor
I can vouch for that a bit too, consuming RSS feeds from academic journals. A regular RSS plugin worked for a while until they added namespace stuff we had to sort through.
#
[tantek]
oh yeah, that too xmlnamespaces--
#
Loqi
xmlnamespaces has -2 karma over the last year
#
[Joe_Crawford]
xmlnamespaces enable the richer format Inkscape uses https://wiki.inkscape.org/wiki/Inkscape-specific_XML_attributes
[KevinMarks] joined the channel
#
[KevinMarks]
I prefer TSV as a replacement for CSV than XML. And yes RSS is bad. I also still maintain a SOAP api
#
[Joe_Crawford]
I am sorry for your necessity to maintain a SOAP API. Such promise. Such a disappointment.
#
to2ds
SOAP? 😱
#
[Joe_Crawford]
How have I never seen that RSS Atom wars page? A depressing ominous delight—history that we must be vigilant to never repeat. https://indieweb.org/RSS_Atom_wars
#
[tantek]
we document it to avoid repeating it
[Murray] joined the channel
#
[Murray]
Just spent a couple of hours spelunking the Wiki looking for prior art for self-hosted locations/venues for check-ins. Surprised to see that most people seem to just link to Foursquare (but I guess Swarm remains a widely used tool). I was specifically looking for URL design pros/cons
#
[Murray]
I'm leaning heavily towards a `location/country/city/street` style pattern, just wondering if anyone knows of any particularly big footguns or dragons in that direction
#
[KevinMarks]
City and street can have varying forms between countries
#
[Murray]
Yeah, not concerned over that, all of the data is normalised by me anyway / follows my own hierarchy 😅
#
[KevinMarks]
The UK is a bit precious about what cobatita city. It also has what are known as postal towns which sometimes get used where the US would put city, which can be confusing
#
[Murray]
Fair, and yeah, I used to work for a postal software company, I'm never going to rely on a third-party database for this kind of thing. Manual 'til it hurts on that front 😉
#
[Murray]
So the data is all already there, I just need to find the structure that will work. Feels like to me that the pattern of `country/city/street` is the most hackable and user friendly, but not sure if I'm missing some obvious/non-obvious downsides
#
[Murray]
(Though your points may explain why no one else seems to have gone down that route)
#
[KevinMarks]
Another UK thing is multiple streets with the same name and overlapping within a postal town, as they're actually in different towns or villages.
#
[KevinMarks]
Here there is Stokesley Road, Marton; Stokesley Road, Nunthorpe and Stokesley Road, Guisborough. However they all have a Postal Town of Middlesbrough, so Google mixes them up.
#
[KevinMarks]
In the Bay Area there is El Camino Real which runs through multiple cities and reuses numbers too
#
[KevinMarks]
So storing a postal code can be a useful disambigution
#
[Joe_Crawford]
like any RESTful interface I think more general to more specific as one reads left to right is best if one isn't going the GUID route.
thegreekgeek joined the channel
#
[mattl]
[Joe_Crawford] Great song... can you tell me if you expand the details, you see text talking about a future delete function? https://libre.fm/user/artlung/scrobble/1729727492
#
[Joe_Crawford]
_This is a scrobble you made, so you should be able to delete it from here. Coming soon._
#
[Joe_Crawford]
[mattl] ^
thegreekgeek_ and jeremy joined the channel
#
[mattl]
Awesome.
#
[Joe_Crawford]
[mattl] Found my first ever scrobble I also put to social media. 2013. https://twitter.com/artlung/status/300658173700874241 but it was passive and pretty noisy.
#
[mattl]
Very nice.