[Lldb-commits] [PATCH] D111409: proposed support for Java interface to Scripting Bridge

Thu Nov 11 06:50:36 PST 2021

Am sorry I can't commit build resources, but am certainly willing to commit time to solving test problems.  In a very general sense, I think our project (Ghidra) will effectively be a test platform for issues with the Java bindings to SWIG.

With regard to the set-up for testing, I have noticed differences in behavior between the "ninja check-lldb-xxx" suite-testing environment, the llvm-lit test sets, and actual lldb runs / individual tests.  I believe (I should never say this out loud, but...) the current code handles these environments correctly.

The Java scripting framework requires the ability to find two things:  the Java install (and within it libjvm) and the SWIG.jar built from the JNI wrappers.  If either LD_LIBRARY_PATH or JAVA_HOME is set in whichever environment is in use, the code should be able to find libjvm.  If CLASSPATH includes the SWIG.jar or we're running out of build, finding it should be handled.  Libjvm is loaded dynamically, so the inclusion of my code "should" have no effect on environments without Java installed.

I was thinking I would also spend a couple of days isolating the swig logic in case supporting the Java interface becomes untenable for you.  Am hoping I can pare it down so that our users could point-and-click at an existing lldb repo, without necessarily having to build llvm/lldb.

Best, Dave

________________________________
From: Raphael Isemann via Phabricator <reviews at reviews.llvm.org>
Sent: Thursday, November 11, 2021 7:23:11 AM
To: David Millar; anoronha at apple.com; fallkrum at yahoo.com; kkleine at redhat.com; medismail.bennani at gmail.com; jonas at devlieghere.com; tedwood at quicinc.com; jmolenda at apple.com; syaghmour at apple.com; jingham at apple.com; vsk at apple.com; boris.ulasevich at gmail.com; lldb-commits at lists.llvm.org; h.imai.833 at nitech.jp; bruce.mitchener at gmail.com; david.spickett at linaro.org; quic_sourabhs at quicinc.com; djordje.todorovic at syrmia.com; serhiy.redko at gmail.com; Liburd1994 at outlook.com
Cc: pavel at labath.sk; mgorny at gentoo.org
Subject: [PATCH] D111409: proposed support for Java interface to Scripting Bridge

teemperor added a comment.

In D111409#3124194 <https://reviews.llvm.org/D111409#3124194>, @labath wrote:

> In D111409#3124075 <https://reviews.llvm.org/D111409#3124075>, @teemperor wrote:
>
>>> Are you asking for dedicated physical resources for running nightly builds?
>>
>> I don't think any of the current bots have a Java installation so I think it's either that or we get someone with a bot to setup the required Java installation.
>
> I don't have a problem with installing the necessary packages on the bot I manage, but I cannot subscribe to tracking down any failures (or flaky tests!) for this configuration (and, in my experience, any new feature like this is going to have flaky tests). Flaky tests (probably just one) are the reason that lua integration is not enabled on this bot.

Sure, I think it should anyway not be up to any bot owner to track down flaky tests. And that nested_sessions lua test is anyway randomly failing everywhere from what I know.

>> FWIW, if no one wants to host a bot for this then I won't mind testing this in own CI <https://ci.teemperor.de>, but I am not using buildbot so we'll have to see if that is acceptable for the community (I could also migrate it to buildbot, but the buildbot interface is just painful to use and I would prefer not to).
>
> I would rather not proliferate test infrastructures.
>
> I'm not sure which pain points are you referring to, but setting up a buildbot instance is a lot simpler these days than it used to be (in particular, you don't need to track down and install any outdated packages).

It's not the setup, it's just that the lab.llvm.org interface is far less usable than Jenkins for tracking down regressions in tests (and Jenkins is already not great, so that says something). But that's just my personal preference and I agree that lab.llvm.org should be the central place for infrastructure.

Anyway, if @labath can run them on his fancy build bot then I would prefer that over having it on my bot (because I pay out of my own pocket and CPU cycles are $$$).

Repository:
  rLLDB LLDB

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D111409/new/

https://reviews.llvm.org/D111409

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20211111/d7fcd287/attachment-0001.html>