DOC: Recreate lost Core/Common example visuals with new CIDs#455
DOC: Recreate lost Core/Common example visuals with new CIDs#455hjmjohnson wants to merge 1 commit into
Conversation
Girder does not support IPFS CIDs. I think it only supports MD5 and SHA512. Searching by filename usually gives results. |
|
As examples used to build without errors, the missing files must be somewhere. data.kitware.com used to be the default for many years, so that should be the first place to look. If search by filename does not yield the (desired) result, trying to download by the old sha512/md5 hash is another avenue. |
|
@dzenanz Both avenues checked — these 20 figures never existed before #440, so there are no old sha512/md5 hashes to try, and a full data.kitware.com filename search (all 16 unique basenames) found no matching content. EvidenceNo prior hashes exist. Girder filename search (per your suggestion).
CID comparison method validated by roundtrip: downloading the published sibling The binaries only ever existed on the #440 contributor's machine; the CIDs were computed locally and the uploads never happened. Happy to restore any figure together with its binary if it ever resurfaces. |
|
Status: the 20 visuals below are confirmed unrecoverable from any public source — the binaries only ever existed on the #440 contributor's machine (CIDs were computed locally; the uploads never happened). Plan: recreate the diagrams, publish the new binaries via an ITKTestingData PR, and convert this PR from reference-removal to The 20 truly missing files(all under The hunt — every source checked
CID verification method (CIDv1/raw/sha2-256) validated by roundtrip on the published sibling file. |
|
The 20 visuals have been recreated as clean conceptual diagrams; binaries are published in InsightSoftwareConsortium/ITKTestingData#63. This PR will be reworked from reference-removal to Review gallery — all 20 recreated figures
|
The 20 PNGs referenced since InsightSoftwareConsortium#440 were never published to ITKTestingData and are unrecoverable from any public source; every build of Module_SphinxExamples fails 20 ExternalData fetches. Point the .cid files at recreated diagrams (ITKTestingData#63) and fix the Operator.png/Operation.png reference mismatch.
hjmjohnson
force-pushed
the
fix/remove-missing-visual-cids
branch
from
5d1bd8e to
36f1ee8
Compare
github-actions
Bot
added
area:Core
hjmjohnson
changed the title
2 participants
Recreate the 20 Core/Common example visuals whose binaries were never published by #440 and are unrecoverable from any public source. The
.cidfiles now point at recreated diagrams (published via InsightSoftwareConsortium/ITKTestingData#63); also fixes theOperator.png→Operation.pngreference mismatch. Blocked on ITKTestingData#63 merging first.Why the original PNGs are unrecoverable — the hunt
gh-pagesCID/<cid>MakePartOfImageTransparent.pngfrom the same #440 IS present)ITK/ITKSphinxExamples/Nightly.sha512/.md5hashesgit log --all --diff-filter=A, any extension.cidin #440; the figures never existed beforeThe binaries only ever existed on the #440 contributor's machine. If the originals resurface they can replace the recreations.
What changed
.cidfiles updated to the CIDs of the recreated diagrams (matplotlib renderings matching each figure's caption/context; review gallery in the comments below and in ITKTestingData#63).ApplyCustomOperationToEachPixelInImage/Documentation.rst:figure:: Operator.pngcorrected toOperation.png— the figure block referenced a name with no corresponding.cidfile whileOperation.png.cidsat orphaned.Documentation.rstfigure references stay live.Local validation
CID computation (CIDv1/raw/sha2-256) validated by roundtrip against the published
MakePartOfImageTransparent.pngblob. Each new blob in ITKTestingData#63 is byte-identical to the file hashed into its.cidreference here.