-
Notifications
You must be signed in to change notification settings - Fork 825
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix schema.rdfa reversions from sdo-deimos/3.0 release #1203
Comments
This (July 29 2015) 55a43a1 was merged into sdo-deimos via #1056 on Mar 24 2016. The commit https://github.com/schemaorg/schemaorg/blob/55a43a126c4981f090036008f49b0b6678f50ef0/data/schema.rdfa is too large to view in Github's Web UI. On commandline "git show 55a43a1" shows a lot of changes. |
hmmm, this |
I do not trace any merge with |
I've committed a very simple diff utility (/cc @mfhepp ) to the repo, https://github.com/schemaorg/schemaorg/blob/master/scripts/differ.py (no commandline flags yet). Here are the outputs, in Turtle syntax, from comparing 2.2 and 3.0 data/schema.rdfa release snapshots (and ignoring extension files):
Looking in these three files for BlogPosting it is clear that the supertype reverted back to Article with the 3.0 release. Since #526 (see http://schema.org/docs/releases.html#g526) in v2.1, and therefore also since 2.2, it should have had a supertype of SocialMediaPosting:
|
I am progressing through the diff file, looking first at triples that were lost between v2.2 and v3.0 definitions. This seems a viable approach, if tedious. Will post a summary of TODOs here in due course. |
https://gist.github.com/danbri/79625878bfbbc1ad098ddd91520aa573 has a first cut at a TODO list based on looking at all triples that went missing from v2.2 to v3.0 and ignoring anything to do with medicine, then investigating the remainder. Next task is to look at non-medical additions that were added in v3 and form a merged TODO list, which is then a basis for a pull request. (aside: I will add the blogposting case too). |
ok. keep me posted to see where I can jump in. |
We have lost also the |
https://gist.github.com/danbri/d79a476cecc1bf425f007524eab11c81 - this is a quick list of terms that were in 2.2 core but not in health-lifesci extension. This is not the best way to check, @RichardWallis is looking into things more deeply but I thought I'd share this since I generated it (with scripts/compare_health.py utility) |
Comparison script added to master 'compareterms.py' Current output from script: https://gist.github.com/RichardWallis/ee96bcfbce20f96e6f967d3caf366a08 Indicates following changes from sdo-phobos (v2.2) to v3.0 master: |
Thanks @RichardWallis! Looking over the summary, I think we can discard half of the "dropped terms" after further investigation. 0.) Missing health-lifesci properties These 4 are not in the 3.0 or extension dirs, and no discussion in issue tracker:
We should consider restoring them, or amending the examples if they are not restored. @twamarc can you comment on any of these 4? 1.) OK - Dropped term http://schema.org/Optometic This became http://schema.org/Optometric 2.) OK - Dropped term http://schema.org/Radiograpy Typo fixed. Now called: http://schema.org/Radiography 3.) OK - Dropped term http://schema.org/specialUsage This became http://schema.org/vehicleSpecialUsage via #705 4.) Dropped term http://www.w3.org/wiki/WebSchemas/SchemaDotOrgSources#source_GoodRelationsProperties Not a vocabulary term. This is an implementation detail for vocabulary acknowledgements, and was changed as part of #1022 |
Here is a slightly reworked summary of "OK" vs "TODO" on all the changes that I found between 2.2 and 3.0. The format is rough but I think I was reasonably thorough, given the scale of the task. https://gist.github.com/danbri/9962d481b1e52025387921cc531a6446 Next task is to strip out the OK items and work up a pull request for the small fixes (reverse unwanted rollbacks) that it lists. |
TODOs
a. TODO: health-lifesci is missing definitions for breastfeedingWarning, healthCondition, prescriptionStatus, secondaryPrevention. Missing: schema:BlogPosting rdfs:subClassOf schema:SocialMediaPosting . 3.) TODO: genre 4.) ingredients - this was already restored. Addressed as a quick fix in http://schema.org/docs/releases.html#g1174 schema:ingredients a rdf:Property ; ... has this gone? seems to be in v3 too. 6.) TODO: check status of manufacturer ... we restored this as a quick fix too, looking at diff of data/schema.org vs data/releases/3.0/schema.rdfa 7.) TODO: netWorth - reversion see 60.) below 9.) TODO: affiliation was subpropertyOf memberOf Established in 2.1 via http://schema.org/docs/releases.html#g596 (Documented that affiliation is a sub-property of memberOf.) 1212 Removed: schema:affiliation rdfs:subPropertyOf schema:memberOf . 2.2: schema:affiliation rdfs:subPropertyOf schema:memberOf . TODO: confirm and restore. 10.) prepTime - TODO: link ISO 2.2: schema:prepTime rdfs:comment "The length of time it takes to prepare the recipe, in ISO 8601 duration format." . 11.) TODO: restore schema:publisher schema:rangeIncludes schema:Person . schema:publisher schema:rangeIncludes schema:Person . has gone. - noted in #1198. 15.) totalTime - TODO link ISO 2.2: schema:totalTime rdfs:comment "The total time it takes to prepare and cook the recipe, in ISO 8601 duration format." . 21.) videoFormat TODO: restore #577 http://schema.org/docs/releases.html#g577 In 2.1 we had http://schema.org/docs/releases.html#g577 "Amended videoFormat to indicate that it is expected on BroadcastEvent and ScreeningEvent, rather than TelevisionStation." TODO: Restore per 2.2. 2.2: schema:videoFormat schema:domainIncludes schema:BroadcastEvent, 27.) parentOrganization - TODO repair and restore. To be restored. Unwanted rollback of http://schema.org/docs/releases.html#g535 ("Broadened domain of parentOrganization to allow any Organization, rather than only LocalBusiness. Noted parentOrganization and subOrganization as inverses.") 2.2: schema:parentOrganization schema:inverseOf schema:subOrganization ; schema:subOrganization schema:inverseOf schema:parentOrganization . ...was also dropped in v3.0. 39.) branchCode TODO 2.2: schema:branchCode schema:domainIncludes schema:LocalBusiness, 45.) cookTime - TODO, restore hyperlink 2.2: schema:cookTime rdfs:comment "The time it takes to actually cook the dish, in ISO 8601 duration format." . 50.) TODO: investigate status of Dentist In 3.0, added: doublecheck addition/edit: 56.) TODO: restore codeSampleType (and sampleType) TODO: We should restore this edit, https://github.com/schemaorg/schemaorg/pull/513/files 2.2: schema:codeSampleType rdfs:comment "What type of code sample: full (compile ready) solution, code snippet, inline code, scripts, template." . TODO: Also fix sampleType (the superseded version). 60.) TODO: netWorth - rollback of #585 We chose to exclude "organization" from the text. This was rolled back in schema.rdfa and should be re-excluded, pending input from FIBO et al. |
Thanks @RichardWallis @danbri , this is helpful.
Therefore I fully agree with TODO: 0) a. |
I propose the following patch (to be included in general pull request):
|
Thanks for the patch, @twamarc . I will work this all up into a larger PR. |
…ical terms to health-lifesci. See #1203 Terms are: * http://schema.org/breastfeedingWarning * http://schema.org/healthCondition * http://schema.org/prescriptionStatus * http://schema.org/secondaryPrevention
Currently missing:
Queued up for re-publication via webschemas.org development site: |
The 3.0 release had this reverted (to Article). See #1203 See also http://schema.org/docs/releases.html#g526
The 3.0 release had this reverted. See #1203 See also http://schema.org/docs/releases.html#g346
The 3.0 release had this reverted, removing HTML entity escape markup. See #1203 Also minor whitespace fixes in RDFa.
The 3.0 release had this reverted, See #1203 #1198 See also http://schema.org/docs/releases.html#g611
…ScreeningEvent (and BroadcastService), rather than TelevisionStation. The 3.0 release had this reverted, See #1203 See also http://schema.org/docs/releases.html#g577
…inverseOf each other. The 3.0 release had this reverted, See #1203 See also http://schema.org/docs/releases.html#g535
Also #1203 as this was (falsely) suspected to be an accidental reversion.
…tions. The 3.0 release had this reverted. See #1203 See also http://schema.org/docs/releases.html#g343 #343
We choose not to mention organizations at this stage. FIBO and other considerations could mean we revisit this, but this commit fixes an accidental reversion that prematurely reintroduced idea that netWorth applied to Organization. If/when we do that, we need to change domainIncludes also. The 3.0 release had this reverted. See #1203 See also http://schema.org/docs/releases.html#g585 #585
Ok this is all queued up for review in #1215 and is pushed to the webschema.org site (but not schema.org itself). Here is a summary of the changes (please change URLs to webschemas.org to review the candidate fixes).
|
For a final check, I suggest we go through http://webschemas.org/docs/releases.html from v2 onwards and double-check that each change still seems to be in place (or is accounted for in this list of fixes). |
This may be over-cautious but for completeness I have gone back in releases.html to v2.0. All the improvements recorded for 2.0 are still current (or have been subsequently evolved further as documented in later releases). 2.0
Site improvements
Quick fixes and Examples
|
Manual review of 2.1, 2.2: 2.2
Site improvements (N/A here) 2.1
Site improvements: (N/A here) Quick fixes and Examples
|
To summarize the manual review from docs/releases.html these are the things that appear to be missing in 3.0 and which #1203/#1215 repair. They are mostly from 2.1. Taking the above list of problems noted in 2.1 section of releases.html, and merging it against the fixes prepared which were based on triple-by-triple data/schema.rdfa comparison of 2.2 vs 3.0, we get the following 10 fixes, each of which is now annotated with the issue # and text of the 2.1 change that is being restored. In addition, this review highlighted that there is a problem with #518: paymentStatus. This fix has been added to the PR and listed here as 0.). 0.)
1.) 2.)
3.)
4.)
5.) 6.)
7.)
8.)
9.)
10.)
|
Ok, as of #1203 (comment) the webschemas.org site is showing the #1215 fixes (i.e. sdo-v3rollbackfixup branch), which amounts to sdo-makemake (3.1 candidate) plus the changes 0-10 above. @RichardWallis can you review these i.e. confirm my belief that this PR #1215 repairs what I say it repairs, and that as far as you can tell it puts us back on track with how v3.0+ ought to be looking? |
I confirm that the changes identified in #1203 (comment) are live on webschemas.org as described. /cc @danbri |
I have merged these fixes into our draft release branch, sdo-makemake. The webschemas.org dev site is currently served from that branch. http://webschemas.org/docs/releases.html#1203 has an overview of the changes. Additional sanity checks welcome. /cc @twamarc |
Announced in https://lists.w3.org/Archives/Public/public-schemaorg/2016Jun/0039.html for wider review. /cc @chaals @shankarnat @pmika @mfhepp @scor @vholland @tmarshbing @rvguha |
I double checked now the health-lifesci.schema.org and I can confirm all terms are in, no missing. |
@danbri I have unfortunately no time for a thorough check, but skimmed over the commits and didn’t notice anything odd. |
Thanks @unor! |
Our last release (sdo-deimos aka 3.0) included a number of unintended reversions. These need careful repair before we open a new branch for a subsequent release. The cause seems to be the March 24 2016 merging of the medical vocab changes into sdo-deimos (#11 (comment)). It seems I (@danbri) did not review the PR with sufficient care and our unit tests did not catch the mistake.
Examples
I can see no simple way to fix this beyond a careful comparison of before/after versions of data/schema.rdfa, which I shall be attempting here.
The text was updated successfully, but these errors were encountered: