Repository navigation
Conversation
Failure markers went to fail/folio, so a file Folio could not thumbnail stayed blank after it learned to, as it does for camera RAW here. They now go to the spec's fail/<program>-<version>, as fail/folio-2: the number is raised whenever Folio learns to draw more.
A thumbnailer that names a type only by an alias never matched the canonical name Folio asks with, so glycin's image/x-dds left a DDS file blank. The declared types are resolved to canonical names once, when the thumbnailers are read.
A CR2, NEF, NRW, ARW, SR2, DNG or PEF showed a blank tile unless a RAW thumbnailer was installed. The camera stores a finished JPEG in each, and a bounded reader of the TIFF directories now finds the largest one. An installed thumbnailer still answers first.
A type with no installed thumbnailer showed a blank tile and nothing else. The details pane now says which optional package draws it, from the types ffmpegthumbnailer, papers, glycin and libgsf declare.
A file nothing installed could draw kept its failure marker, so a thumbnailer installed afterwards never ran for it. The marker now records the thumbnailer the type had, none included, and holds the file back only while that is unchanged.
One orientation was kept for the whole file, so a preview whose directory names a turn of its own was turned by IFD0's instead. Each embedded JPEG now carries its directory's orientation, and takes the file's only when its directory names none.
With Poppler installed and no PDF thumbnailer, a PDF that failed to draw (damaged, or locked with a password) had the details pane name papers as missing. A PDF's package is now named only when pdftoppm is not there either.
The list and the columns asked for a thumbnail only for a photo or a video, so a PDF, a song with a cover or a document had its picture in the grid and an icon in a row. A row now asks for every picture a tile shows but a text file's page, which says nothing at a row's size, and a page or a cover is fitted whole in the row where a photo is cropped.
A thumbnailer's command was filled in one code after another, so a file called 50%off.mp4 had the %o in its own name replaced by the output path. The thumbnailer was sent a file that is not there, and the failure was remembered. Each code is now read once.
A failure marker keeps every app from trying a file again, and one was written for a thumbnailer that never started, for a spec whose program is not installed, for a file that could not be read and for one moved away or written to during the run. None of those says the file is broken, so none is remembered now, and a PDF whose thumbnailer cannot start is left to pdftoppm.
A thumbnailer still running after 15 seconds was killed alone, so the programs it had started (glycin and the office thumbnailer start their own) ran on. Each now runs in a session of its own, under a keeper that stops the whole session past its time, and when Folio itself is gone.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Whether a failed run is remembered now reads as one condition. The new thumbnailer tests use the ended() helper main already has, which counts a zombie as ended, in place of a second copy of it. A .thumbnailer file is called that rather than a spec, as on the RAW branch. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…er thumbnails Folio waits for the thumbnails being made as it quits, so a hung thumbnailer kept it running in the background for up to 15 s. Thumbnails now has quit(), called on aboutToQuit, and run() waits in short slices so it stops the helper at once; a file stopped that way is not marked as failed. Start and finish share one deadline instead of each getting the full patience. A thumbnailer that exits while something it started in the background still runs no longer leaves that behind: its session is killed after every run, not only after a timeout. Rows ask for 128 px and tiles for 256 or 512, so a folder seen as tiles and then as rows ran every thumbnailer twice. load() now scales down a fresh thumbnail from a larger bucket before making one. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
[Medium impact] The PR appears safe to merge; no new actionable issue was found. SummaryRows now request the same thumbnails as tiles, except text pages. Thumbnail commands expand percent codes once, and failure markers are saved only after a file was judged.
|
…n Folio quits Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This was referenced Nov 13, 2026
This was referenced Nov 18, 2026
Contributor
Author
|
Thank you @thisisgm, this was careful work. Its "rows show pictures" part already landed as #149, and the rest has been taken into focused pull requests off today's main, each crediting you:
What isn't carried on: the stop-on-quit and keeper-after-Folio-dies machinery, reusing a larger cached bucket, and the RAW branch merges (those are handled from #66). Closing this in favor of those PRs. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
By GM (@thisisgm), from #50: the work and the description below are theirs, with their commits kept as they made them. It's stacked on #66 (their #17 with its review fixes), and this pull request carries it on from our own branch, with the fixes from review added on top.
Four changes to thumbnails, one commit each with its own tests: a row in the list and the columns shows the picture its tile shows, a file whose name holds a thumbnailer code reaches the thumbnailer as it is, a failure is remembered only when the file itself was judged, and a thumbnailer past its time is stopped together with what it started. No library,
dependsoroptdependsentry is added.What you'd notice
50%off.mp4gets its thumbnail. A thumbnailer'sExecline was filled in one code after another, so the%oin the file's own name was replaced by the output path. The thumbnailer was sent a file that is not there and the failure was remembered. Each code is now read once, in one pass, and%%is a percent sign as the spec says.Execprogram is gone is no longer taken as a thumbnailer, so the details pane still names the package that would draw the type, and a PDF whose thumbnailer cannot start is left topdftoppm.Numbers
Measured on a mini PC, cold start, six runs a view, on a folder of 2,000 pictures and videos. These runs are from the rows commit before the rebase onto the current main, against the main of that day. The keeper adds one fork to each thumbnailer run; that is not in these numbers yet.
A folder of photos and videos asks for the same thumbnails as before, and the ranges overlap in every view. A folder of PDFs or music now draws its rows' pictures, which costs what the grid already paid for them.
Screenshots
The same folder in the list, main on the left and this branch on the right: the song shows its cover and the PDF its first page.
Changed in review
Reviewed with Codex.
🤖 Generated with Claude Code