[
https://issues.apache.org/jira/browse/XALANJ-2651?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18012925#comment-18012925
]
Joe Kesselman edited comment on XALANJ-2651 at 8/8/25 9:45 PM:
---------------------------------------------------------------
+_*xalan-test really was supposed to be a cross-platform XSLT conformance test,
just as other language-standard test suites are.*_+ Despite the name, it was
*not* intended to only be Xalan's regression tests.
As such, it shouldn't be surprising that it's a separate toolkit, any more than
any other reusable toolkit is. Or that it is a separate development project.
Even if most of the expertise (and just about all of the actual support) is
shared.
+_*Git merge-request validation is*_ _*already running these tests.*_+ They
just fetch the tests and driver explicitly from the xalan-test project. See
.github/workflows/main.yml. They do not run all the tests, because at this time
not all of the tests run as cleanly as one might want; we need to update the
distinction between known failures being worked, known failures that Xalan has
accepted as implementation limitations, and tests which Xalan expects to
succeed – with separate sets for interpretive and compiled mode, alas. But they
run the core set. +_*So that isn't particularly strong justification for
reorg.*_+
+_*We could reasonably move the Xalan-specific tests into the Xalan-J*_
*project.*+ The nonstandardized extension tests, and some of the API tests, are
certainly Xalan-unique. The question is whether/how drivers need to be
duplicated, or if they should be copied/moved into xalan-java/test. On the
other hand, that might slightly hinder compatibility testing between Xalan and
other processors.
+_*It might be more Mavenish to have drivers living in xalan-java/test,
but...*_+ They'd still want to reference xalan-test for the test suite. And
after the battle I had to get Maven to do builds at all "properly", I am
frankly not especially excited about another round of "The Maven Way just
because it's The Maven Way."
{+}_*I HIGHLY recommend opening a separate work item for xalan-test
rationalization and/or xalan-java/test reorganization*_{+}. The former is
probably overdue. The latter strikes me as a much lower priority than most of
what's hanging on our queue.
+_*I do not accept either as part of this work item.*_+ The Maven port is
massively overdue, and I want it +_*CLOSED OUT*_+ so I can go back to working
on Xalan itself.
+_*New suggestions like this should go in new work items, to be discussied,
prioritized, and worked on separately.*_+ If someone wants it more urgently, I
cordially invite them to claim that work item and make it happen. If it isn't
breakage due to this work item, or otherwise necessary to complete this work
item, it is +_*not part of this work item.*_+
Yes, I do mean to be that emphatic. I am not really happy with the Maven
migration as a whole; it may have been necessary in some sense, but it cost
far more than was justified. Let's do what remains absolutely necessary and
call it cooked so I can work on things that actually matter. If someone else
wants to polish it further, great; assign yourself the task like any other.
was (Author: JIRAUSER285361):
+_*xalan-test really was supposed to be a cross-platform XSLT conformance test,
just as other language-standard test suites are.*_+ Despite the name, it was
*not* intended to only be Xalan's regression tests.
As such, it shouldn't be surprising that it's a separate toolkit, any more than
any other reusable toolkit is. Or that it is a separate development project.
Even if most of the expertise (and just about all of the actual support) is
shared.
+_*Git merge-request validation is*_ _*already running these tests.*_+ They
just fetch the tests and driver explicitly from the xalan-test project. See
.github/workflows/main.yml. They do not run all the tests, because at this time
not all of the tests run as cleanly as one might want; we need to update the
distinction between known failures being worked, known failures that Xalan has
accepted as implementation limitations, and tests which Xalan expects to
succeed – with separate sets for interpretive and compiled mode, alas. But they
run the core set. +_*So that isn't particularly strong justification for
reorg.*_+
+_*We could reasonably move the Xalan-specific tests into the Xalan-J*_
*project.*+ The nonstandardized extension tests, and some of the API tests, are
certainly Xalan-unique. The question is whether/how drivers need to be
duplicated, or if they should be copied/moved into xalan-java/test. On the
other hand, that might slightly hinder compatibility testing between Xalan and
other processors.
+_*It might be more Mavenish to have drivers living in xalan-java/test,
but...*_+ They'd still want to reference xalan-test for the test suite. And
after the battle I had to get Maven to do builds at all "properly", I am
frankly not especially excited about another round of "The Maven Way just
because it's The Maven Way."
{+}_*I HIGHLY recommend opening a separate work item for xalan-test
rationalization and/or xalan-java/test reorganization*_{+}. The former is
probably overdue. The latter strikes me as a much lower priority than most of
what's hanging on our queue.
+_*I do not accept either as part of this work item.*_+ The Maven port is
massively overdue, and I want it {*}{*}{+}_*CLOSED OUT*_{+} so I can go back to
working on Xalan itself.
+_*New suggestions like this should go in new work items, to be discussied,
prioritized, and worked on separately.*_+ If someone wants it more urgently, I
cordially invite them to claim that work item and make it happen. If it isn't
breakage due to this work item, or otherwise necessary to complete this work
item, it is +_*not part of this work item.*_+
Yes, I do mean to be that emphatic. I am not really happy with the Maven
migration as a whole; it may have been necessary in some sense, but it cost
far more than was justified. Let's do what remains absolutely necessary and
call it cooked so I can work on things that actually matter. If someone else
wants to polish it further, great; assign yourself the task like any other.
> Migrate build from Ant to Maven
> -------------------------------
>
> Key: XALANJ-2651
> URL: https://issues.apache.org/jira/browse/XALANJ-2651
> Project: XalanJ2
> Issue Type: Improvement
> Security Level: No security risk; visible to anyone(Ordinary problems in
> Xalan projects. Anybody can view the issue.)
> Components: Build
> Reporter: Gary D. Gregory
> Assignee: Joe Kesselman
> Priority: Major
>
> Migrate the build from Ant to Maven.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@xalan.apache.org
For additional commands, e-mail: dev-h...@xalan.apache.org
