- From: Rob Stewart <robstewart57@gmail.com>
- Date: Wed, 6 Jan 2016 00:44:11 +0000
- To: Gregg Kellogg <gregg@greggkellogg.net>
- Cc: RDF Comments <public-rdf-comments@w3.org>, public-rdf-tests@w3.org, "Eric Prud'hommeaux" <eric@w3.org>
- Message-ID: <CA+ZOasJ81Xg2xe5LwNOj2JKGW5AbFff3kW8AgiNEFsE_SMSePg@mail.gmail.com>
On 5 January 2016 at 23:35, Gregg Kellogg <gregg@greggkellogg.net> wrote: Personally, I’ve had problems with git sub-modules before in that they get > stuck on the commit they’re initialized with, but, in principle, this is > probably the way to go. Of course, you can also simply access them over > HTTP. > I've gone ahead and done this. https://github.com/robstewart57/rdf4h The rdf-tests repo is a submodule, and as you say it's pinned at commit 67a7ab9. It'd be up to me update and pin a newer rdf-tests commit hash, though I've documented this in the README: > If you've never initialised the rdf-tests repository: > > $ git submodule update --init --recursive > > To run the parser tests against the latest W3C test files: > > $ git submodule foreach git pull origin gh-pages And I've set TravisCI up to pull the latest from w3c/rdf-tests before running the testsuite by modifying the .travis.yml file: before_install: - git submodule foreach git pull origin gh-pages Which seems to be working. I guess there's two possible advantages of using git submodules for staying in sync with w3c/rdf-tests rather than using wget to get GitHub or apache tar files. 1) when parser tests fail, I'll be able to pinpoint the exact w3c/rdf-tests commit that the test fails against, and 2) `git submodule foreach git pull origin gh-pages` will only pull newer files, rather than downloading all files in a tar every time. The setup seems reasonably robust, but if using w3c/rdf-tests as a submodule to synchronise is more flaky than I'd have hoped, I'll report back. -- Rob > On 4 January 2016 at 16:56, Gregg Kellogg <gregg@greggkellogg.net> wrote: > >> Eric and I discussed this further yesterday, and came to some different >> conclusions: >> >> There is value in reusing the existing test suite URIs; the need to >> preserve the original state of these test suites is served by keeping a >> tarball of the original state of the suites. >> >> There is precedent for updating the implementation reports in place, as >> this has been done already for Turtle, TriG and a couple of others. They >> can maintain a reference to the version used for the transition call >> (although this seems broken in at least one case). >> >> A post-commit hook can easily be used to synchronize the content of the >> RDF Test group’s Github repository in W3C spec, thus guarding against some >> future change that might make it unavailable remotely. >> >> Eric is going to experiment with this. He’ll create tarballs of the >> existing directories, which can be added to our Github repo. Once we’re >> satisfied, the existing test directories and implementation reports can be >> set to either redirect to the synchronized location, or Apache >> configuration can be updated to make these appear in the existing locations. >> >> This can help avoid confusion about what the “official” tests are. A >> future WG may, of course, change this, but is more likely to create a new >> location for tests, and certainly implementation reports. >> >> Gregg Kellogg >> gregg@greggkellogg.net >> >> On Jan 3, 2016, at 12:40 PM, Gregg Kellogg <gregg@greggkellogg.net> >> wrote: >> >> Hi Rob, the group discussed this on RDF Tests [1] and came to the >> conclusion that leaving the existing tests in place would be better, with a >> reference to the new tests and implementation reports. Of course, we may >> revisit this in the future, but the CG has no official standing, and there >> are no active RDF nor SPARQL working groups. >> >> I prepared alternate landing pages for each of the test suites and >> implementation reports to refer to the new locations [2][3]. I asked Eric >> in November to replace the original pages with those referenced, but that >> obviously hasn’t happened yet. Eric, can we make this happen? >> >> Gregg Kellogg >> gregg@greggkellogg.net >> >> On Jan 3, 2016, at 3:54 AM, Rob Stewart <robstewart57@gmail.com> wrote: >> >> Hi, >> >> The following URLs are each a pair of W3C RDF parser tests on the w3.org >> page and the GitHub URL for the corresponding directory. The files in the >> w3.org directories don't appear to be up to date with the GitHub >> directories. Is the w3.org directories going to be periodically updated >> with the GitHub files, or should the w3.org directories be treated as an >> archive? >> >> http://www.w3.org/2013/TurtleTests/ >> https://github.com/w3c/rdf-tests/tree/gh-pages/turtle >> >> http://www.w3.org/2013/N-TriplesTests/ >> https://github.com/w3c/rdf-tests/tree/gh-pages/ntriples >> >> http://www.w3.org/2013/NQuadsTests/ >> https://github.com/w3c/rdf-tests/tree/gh-pages/nquads >> >> http://www.w3.org/2013/RDFXMLTests/ >> https://github.com/w3c/rdf-tests/tree/gh-pages/rdf-xml >> >> http://www.w3.org/2013/TriGTests/ >> https://github.com/w3c/rdf-tests/tree/gh-pages/trig >> >> Thanks, >> >> -- >> Rob Stewart >> >> >> >> Gregg Kellogg >> Chair RDF Test Curation CG >> >> [1] >> https://lists.w3.org/Archives/Public/public-rdf-tests/2015Nov/0006.html >> [2] >> https://lists.w3.org/Archives/Public/public-rdf-tests/2015Nov/0016.html >> [3] >> https://lists.w3.org/Archives/Public/public-rdf-tests/2015Nov/0018.html >> >> >> > >
Received on Wednesday, 6 January 2016 00:45:01 UTC