Ergonomic, hermetic, user-friendly software build system for Python, Go, Java, Kotlin, Protobuf, Scala, Shell. Pants lets you fearlessly scale up your codebase!

Pants

TIL: GitHub has a URL shortener <https://github.blog/2011-11-10-git-io-github-url-shortener/>

<https://twitter.com/scienceshitpost/status/1220839553662996480?s=20>

TIL: regex101 can help you visually confirm that your regular expression does what you think it does (for instance, <https://regex101.com/r/SUZCl6/3>). Thanks <@UBV8BCQPJ> for the tip :smile:

does the `dep-usage.jvm` task requires OpenJDK 8?

its looking for `tools.jar` which is gone since JDK 9

cc <@U066LC126>. I think you’ve been looking into this problem?

I’m not familiar with the dep-usage case, but in the javadoc case, the issue is that javadoc can no longer be used as a jar--you have to use the binary

I think it’s worth filing an issue specific to dep-usage over

<https://github.com/pantsbuild/pants/blob/master/src/python/pants/backend/jvm/tasks/analysis_extraction.py#L155>

I just saw the tool mentioned in the docs today, and wanted to report it didnt work anymore

will need to use the binary rather than invoking java

TIL `Revision.lenient("1") &gt; Revision.lenient("1.0")`

what is `Revision.lenient("1") &gt; Revision.lenient("1.1")`?

pants’ notion of what a revision is has some bugs

it’s used for things like deciding which jdk to use

it only is now becoming an issue because you can specify eg version 9

```                             self.assertTrue(Revision.lenient("1") != Revision.lenient("1.0.0"))
                             self.assertTrue(Revision.lenient("1") &gt; Revision.lenient("1.0.0"))
                     &gt;       self.assertTrue(Revision.lenient("1") &gt; Revision.lenient("1.1.1"))
                     E       AssertionError: False is not tru```

the issue is that if you request version 9 as the min version, version 9.0 will be too small

found the bug. Also, I discovered a new thing I don’t like about python--but once I found the right docs it made sense.

I can think of two behaviors that would make sense.

arguably 1 and 1.0 shouldn’t be equal, since they are different values. That argues in favor of `1 &lt; 1.0(.0)*`

because asking for `1` should allow `1.0`

I wrote a thing for comparing semvers a couple years ago. I believe the standard for comparing equality is to pad all the shorter values with as many `.0`s as necessary

that implies that I should check eq more complexly as well

(The thing I wrote was specifically for semvers as used by maven, which has all kinds of additional complexity)

currently it’s `components == components`

If you said “i need version 1.0” would you be satisfied by version 1? 