jmeter-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Vladimir Sitnikov <>
Subject Re: Release JMeter 5.2 ?
Date Sat, 05 Oct 2019 07:40:55 GMT
>which means for now there is a
>regression on Thread group that is not fixed

Unfortunately, that commit did not really fix the regression.

Here's a test to validate test plan loading:

>Do you intend to merge the facelift of Thread Group so that I can commit
>the fix?

I do intend, however current implementation breaks Visualizers and,
probably it breaks JMeterPlugins.

Here's a sample:
[image: Снимок экрана 2019-10-05 в 10.36.47.png]

What happens here is AbstractVisualizer adds "write results to file" right
into the "titlepanel", and it is appended to "the first column".

I think we need a procedure to automatically generate screenshots.
That would help with keeping documentation up to date, and it would help to
review UI-related PRs.

>How do you want to proceed so that we don’t « walk on each other toes » ?

I suggest we follow "commit then review process".
However, I guess we could ask each other if anybody else is working on the
same thing.
For instance, Graham did ask if anybody else was working on JUnit5
In the same sense I happen to refrain from merging Spock-related changes
because I don't actively use Spock, so it is harder for me to judge.

It should be OK to revert changes.

The fix of "same user on each iteration" had two changes (original
implementation, and the recent fix).
Unfortunately, none of the two included tests regarding backward
I see how it could be tempting to just commit a 2-3line long fix and close
the subject, however, it was extremely annoying for me to see
the commit that is both "conflicting with PR changes" and "which does not
fix the issue".

Of course, I could have behaved in a better, but, well.
The feature broke backward compatibility in a hidden from user way.
The tests are missing.
Let us avoid the same things in the future?

As I am developing "thread group UI update", I need to test id. Literally.
I create test plans. I load old jmx files, I change window size, and so on.
"same user on each iteration" bug does impact user experience up to the
point when I decided to include the fix right into UI update PR

The test is really not that complicated to write:

PS. You might want to review


  • Unnamed multipart/related (inline, None, 0 bytes)
View raw message