#[jeremycherfas]I've been impressed and pleased by Photo Mechanic’s reverse geocoding tool for phone photos. Adding latlong to “proper” photos using Compass would be cool if I could integrate it easily. Good project to work on.
mro, shaunix[d], justAstache, benji, kloenk, KranNaut and tetov-irc joined the channel
#GWGI'm not sure how I envision such a script working.
[jgmac1106], jacky, mro, Ramon[d], whatisthis, sayanarijit[d], hepphepp[d], ancarda, capjamesg, srushe, [tantek] and [schmarty] joined the channel
#Loqi[jasonbosco] Phrase searching using quotes is available in `0.23.0.rc42`.
If you replace the version numbers on the [downloads page](https://typesense.org/downloads/) with this RC version number, you should be able to get the Docker Image or DEB package.
...
#barnabySeirdy: getting srcset parsing support implemented could be part of a solution to that, but IMO most mf2 consumers should be fetching images and scaling them as appropriate
#Seirdyso perhaps yet another recommendation to make sure photos look good when scaled up/down to a variety of resolutions, from 32x32 to 512x512
#barnabyyep, or using srcset to suggest different images for use at different sizes
#Seirdywell we shouldn't recommend using <img srcset> or <picture> until the mf2-parsing spec incorporates them and an implementation or two rolls out.
#Seirdyuntil then, we should recommend using a single image that looks good at a variety of sizes.
#[tantek]or better yet a hybrid approach, use whatever img/srcset/picture markup gives your visitors a better experience on your site, and then provide the "best you can do today" microformats markup within that
#[tantek]and then we can look at that gap between what site viewers experience, vs the "downlevel" data from the mf2 of the page, and analyze it with the actual img/srcset/picture markup to see what the potential could be if more/all of that srcset/picture markup were parsed into the mf2 results
#barnabyyeah exactly, using srcset/picture to improve page weight now has immediate positive gains, and is future-proof for whenever we decide on and implement mf2 parsing support
#[KevinMarks]You can do srcset with img fallback I think
#Seirdy[KevinMarks]: srcset is part of img already
#[KevinMarks]Right, so you have a src and a srcset so the mf2 parse ok
#barnabyyep, I do that on my site. <img> with src and srcset attributes
#Seirdycombining <img srcset> and <picture> is where things get tricky
#barnabyin what circumstances would you combine the two?
#Seirdyif each <source> in the <picture> has srcsets and sources vary depending on media queries for prefers-color-scheme and supported image formats, you could easily end up with 12+ different image files for a single <picture> element.
#Seirdysrcsets are suggestions while <source media=""> is a mandate.
#Seirdymy personal philosophy is "make the image as small as you can until the information you wanted to convey would be lost by further shrinkage".
#Seirdyif you can, just srcsets for resolution and <picture><source>...</picture> for media queries unrelated to resolution or viewport size, unless you *know* exactly what the right image to pick is for each given resolution+DPR.
#Seirdyin which case you can use <picture> for everything
#barnabyhuh I somehow missed the sizes attribute when I was getting srcset working on my site. Firefox seems to be able to pick the correct image to use anyway, based on my testing
#barnabyI get why <picture> can be useful, but I’ve never had any reason to use it myself
#barnabywould a mf2 parsing algorithm for srcset need to support the sizes attribute too?
#SeirdyTLDR: srcset is for different resolutions for a given image, and <picture>+<source> is for different images altogether. <picture> is good for using next-gen formats and a dark/light/high-contrast/low-contrast variant
#barnabyit seems to be more specific to the way the image is presented on the original site, which might not apply to re-use
#barnabySeirdy: by different resolutions, are you talking about different sizes, or strictly different DPI?
#Seirdyi'd assume that the client should "know" what sizes to look for ahead-of-time
#Seirdyif i wanted i could make it fully automatic using hugo's image processing but i like to make sure each image is as small as possible at the minimum acceptable quality.
#barnabyah yeah that’s an advantage of static site workflows
#barnabywonder how long it’ll take for imagick to support these new formats
#SeirdyWebP2 will obsolete both AVIF and WebP. AVIF is such a broken image format; its lossless mode is completely useless and is typically beaten by WebP or even PNG+Zopfli, since its encoders are designed for lossy compression of animated content. WebP2 incorporates AV1's lossy algos but uses its own lossless algos, just like WebP did with vp8 (lossy) + its own lossless format.
#barnabyand for current versions of imagick to become available on shared web hosts
#Seirdybarnaby: probably not too long since these encoders are built with good C APIs
#SeirdyJXL obsoletes WebP but not AVIF, since the lossy compression works quite differently
#SeirdyAVIF removes detail while JXL introduces artifacts
#Seirdywhich makes them better suited for different tasks
#Seirdyessentially, in the distant future: for lossless compression, make a JXL and WebP2 and pick the smaller one; for lossy compression, decide if you want to remove detail or introduce artifacts, then pick WebP2 or JXL acordingly. then make a PNG or JPG to support legacy clients. Repeat for dark mode if your image is really bright. If your image is huge, repeat for a smaller size and use srcsets
#barnabycalling GIF a “video format” is a bit of a stretch!
#Seirdyit's your only option for royalty free raster animation on Fruit Company mobile devices
#barnabybrings back memories of my first ever (pre-youtube) website, where I tried to upload five minute long films as animated GIFs because the website builder didn’t support movie files
#Seirdyno theora, no vp8, no vp9, no av1. they just have "experimental" vp9 decoding support that they only enable for WebRTC
#Seirdybarnaby: the only overview you need is "JXL or WebP2 tomorrow; AVIF and WebP today." atl in the context of web browsers
#Seirdyand always use a png/jpg fallback for legacy clients
#SeirdyJP2 died before it lived, and HEIC is patent encumbered with royalties making it DOA for Web content (or for anywhere on my machine if I can help it)
#SeirdyWebP isn't exactly a "next-gen" format tho it isn't a "legacy" format either. Even Safari on iOS supports WebP, including animated webp. lossless animated webp honestly may never be dethroned
#Seirdywithout a moving cursor, mpdecimate removes almost all frames leaving the result at 4-8 fps.
#Seirdylossless animated webp is perfect for this; it's even smaller than the AV1 video encoded with libaom.
#barnabyfirefox seems to have some UX issues with it unfortunately, it played once and I can’t rewind it. it’s stuck on the last frame even when I refresh the page
#Seirdybarnaby: it's an animated image format, and i don't think it loops without an html attribute.
#barnabyon my machine it’s handled super weirdly, even refreshing while its playing doesn’t interrupt playback, and opening it in a new tab loads it at the same frame as the previous tab
#Seirdywebkit's javascriptcore is fine, it's webcore that's a mess.
#barnabyugh, time to report more FF bugs I suppose
#Seirdybarnaby: this could be an issue with a distro package
#Seirdytry it in an official build, those use vendored deps and a different toolchain.
#barnabyI’m on mac os, so I don’t think I have much choice in that regard?
#Seirdyoh i forgot non-linux/bsd operating systems existed