Media 2022 queries

[eeeps]

Progress on #2884

Here's our todo list:

Images

• [x] Single-pixel images (image_1x1.sql)
• [x] What percentage of pages include an <img> element? (at_least_one_img.sql)
• [x] What percentage of pages make a request for an image resource (at_least_one_image_request.sql)
• [x] What percentage of pages have an image or background image as the LCP element? (lcp_element_data.sql)
• [ ] Most common colors
• [ ] Most common colors by popular (and geographically distributed) set of TLDs
• [ ] Ranked list of color spaces
• [ ] Percentage of image resources that are animated
• [ ] Distribution of frames per animated image
• [ ] Prevalence of ITPC copyright
• [ ] Prevalence of EXIF thumbnails
• [ ] Time series of format adoption over time (or a simple YoY comparison of percentages)
• [ ] Distribution of pixel counts
• [ ] Distribution of pixel counts excluding single pixel images
• [ ] Ranked list of aspect ratios
• [ ] Distribution of bytesizes
• [ ] Distribution of image bits/pixel
• [ ] Median bits/pixel/frame by format
• [ ] Most common metadata fields
• [ ] HTML vs CSS vs Javascript vs favicons - counts per page
• [ ] slice bytesizes, pixel counts, formats by embedding context, too
• [ ] Largest favicon
• [x] Percentage of <img> elements that are lazy-loaded (over time) (lazy_loading_adoption_over_time.sql)
• [x] Percentage of lazy-loaded <img> elements that were LCP (lcp_element_data.sql)
• [ ] Percentage of images with an alt or figcaption
  • [ ] weed out unuseful ones
• [x] Percentage of srcsets containing x descriptors (image_srcset_descriptor.sql)
• [x] Percentage of srcsets containing w descriptors (image_srcset_descriptor.sql)
• [ ] Distribution of number of candidates in srcsets
• [x] Distribution of image densities (image_srcset_densities.sql)
• [ ] Usefulness of min->max desnity ranges
• [ ] Distribution of w descriptor errors
• [ ] Distribution of sizes errors
• [ ] Impact of inaccurate sizes - pixels
• [ ] Impact of innacurate sizes - bytes
• [x] Most common sizes values (image_sizes_attribute_strings.sql)
• [x] Percentage of <picture> elements using type (picture_switching.sql)
• [x] Percentage of <picture> elements using media (picture_switching.sql)
• [ ] Most common media queries in media attributes
• [ ] Number of candidates in type attributes
• [ ] Most common type stacks
• [ ] Intrinsic vs extrinsic vs both sizing
• [ ] Percentage of extrinsic-in-one-dimension images that have height and width attributes
• [x] Cross-origin vs same-origin (img_xdomain.sql) TODO: domain vs origin...

Video

• [x] percentage of pages including a <video> element
• [x] Distribution of video durations
• [x] Ranked list of most-common formats
• [x] Percentage splits of CSS display values of <video> elements
• [x] Ranked list of <video> element attributes
• [x] Ranked list of <video> preload values
• [x] Ranked list of <video> autoplay values
• [x] Ranked list of <video> width values
• [x] Split of <video><source> vs <video src>
• [x] Distribution of number of <source> children
• [x] Most common type stacks

[eeeps]

I've copied over our 2021 queries, which will check off a lot of these boxes. Will reconcile soon!

[siakaramalegos]

@eeeps @akshay-ranganath How are things going with the queries. It looks like you have 57 but the checkboxes are still mostly unchecked so I can't tell. Do you think you can wrap up the queries and have the data and charts in your results sheet by the end of the month? We want to keep the writing on track for publication. Thanks!

[siakaramalegos]

Hi @eeeps @akshay-ranganath can you give me a written description of where you're at/how it's going? See my last comment. We're trying to understand progress on all the chapters. Thanks

[eeeps]

@siakaramalegos Plan A was to get all of the queries written by the end of July; going a bit too far down a couple of rabbit holes regarding the exciting new data from ImageMagick and EXIFTool, plus me not planning for my end-of-month vacation, foiled that plan.

So now we're on to Plan B, with a firm deadline of next Monday the 8th to get all of the queries written and results in, which will leave 3 weeks for writing and editing. Likely we will have to simplify a few queries (e.g., just querying named ICC profiles, rather than trying to figure out every way every format can possibly encode non-sRGB profiles + how EXIFTool reports on them all...), but having a write-about-able set of results in a week's time seems accomplishable.

Thanks for checking in.

[eeeps]

Oh no I made a mess

[eeeps]

Ok I think I cleaned the mess up 😬

[rviscomi]

@eeeps there are a couple of linter errors, otherwise is this PR ready for review?

[rviscomi]

🛎️ @eeeps @akshay-ranganath can we wrap up the remaining queries and mark the PR ready for review? The deadline for having the chapter written, reviewed, and edited is September 1, so it'd be great to finish the analysis and start writing ASAP.

[eeeps]

Still todo:

1. Make the linter happy
2. Finish an Animated GIF query that's very close...
3. BONUS: Leave some notes for future analysts, probably in a comment on this PR

I'd planned on buttoning this up yesterday but it's stretched into today. Today for sure.

[rviscomi]

BONUS: Leave some notes for future analysts, probably in a comment on this PR

I'd suggest creating something like a "Notes for 2023" section in the README

[eeeps]

I see that there are some other queries that we haven't got results for in the sheet. I'll straighten all of that out. But with that last Animated Gif query, the queries themselves are, indeed, ready for review.