Confirmed that the Maven-generated doc includes all the right files, and
is within epsilon of being commit-ready.
Some of them are being output in somewhat different places. All the
results are currently going into target/site/xalan/ rather than there
being a separate xml-site/target/xalan-j output tree. That can be
juggled; not my highest priority right now.
Some of those documents are going into target/site/xalan/local/. I'm not
sure why. Again, not my top priority right now.
Some .gifs are being put into target/site/xsltc/images rather than
build/xalan-j_*/xdocs/sources/xsltc/. Frankly, easiest fix here may
just be to fix the references from the generated HTML.
-=-=-=-=-
I *think* my next step is to run a full regression suite on the Maven
build, comparing test results with those from an Ant build, to make sure
the code's all being properly copied/built/linked and that my
substitution of maven-provided binaries hasn't broken anything.
Assuming that passes, we can discuss whether to merge now and make final
documentation cleanup a separate work item.
I'll start by testing on Linux. If someone else is willing to to do the
"run all the tests for both builds and compare results" for Windows
and/or Mac, I'd welcome the assistance. If not, I'll get to Windows later.
-=-=-=-=-
If there's anything else that people particularly want me to check
before issuing a pull request, please let me know.
On 10/5/2023 6:08 PM, Joseph Kessselman wrote:
Ported my .sh files, creating a mvnbuild.bat for use on Windows, along
with the supporting stylebook_docgen.bat (since I haven't yet moved the
stylebook processing into the pom.xml).
Running these on Win10 with AdoptOpenJDK's jdk-8.0.362.09-hotspot builds
without errors.
***BUT*** I haven't yet properly (re)tested the generated artifacts on
either Linux or Windows
***AND*** I haven't yet walked through all of the generated
documentation contents to compare them with what the Ant build produced
and make sure things are landing in the proper (relative) locations.
***AND*** I don't have a Mac available for testing on that platform.
This is where multiplatform Continuous Integration would be nice to
have... assuming it does have all the VM environments we'd want to test
under.
So basically I'm at the sanity-checking stage of the process.
Other things to be considered:
Squashing the manymany commits on this branch before issuing a pull
request.
Deciding whether to back out some or all of the javadoc fixes I put in
before I figured out how to turn off doclint. Technically they're a
distraction from this migration effort, and they're quick-and-dirty, and
as I've said there are probably better ways to handle our javadoc
generally... but I'm reluctant to make the effort to back them all out
again. Opinions welcome.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@xalan.apache.org
For additional commands, e-mail: dev-h...@xalan.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@xalan.apache.org
For additional commands, e-mail: dev-h...@xalan.apache.org
