David M. Lloyd
David M. Lloyd
In `ReaugmentTask`, there is long-standing code to force the packaging (and disable native), which I would have expected to take effect in this case. But I don't see any equivalent...
Last time I saw this issue (which was recently, but yet I don't remember where it was), it was because of a class which had `javax` vs `jakarta` (i.e. a...
I think that would make sense.
I believe the first step would likely be adding support to JVMCI.
I haven't found any documentation on the JVMCI side, but I believe that 32-bit ARM (in normal operation) uses 16 32-bit registers, one of which (R15) is the PC and...
I posted this in the other thread: > I believe that custom qualifiers come after predefined qualifiers (like `SP`) according to the Maven rules. > > ``` > $ jbang...
Well, that would be nice. :-) However, Maven doesn't have any way to know whether you intend to put `Final`/`0`/etc. before a qualifier. It only has rules for leading zeros...
To be perfectly clear, my `jbang` examples above were also from the Maven resolver, same class. So we agree that Maven presently considers `3.8.5.SP1 < 3.8.5.redhat-00005`, and the SmallRye version...
Given everything, I'm going to call this "not a bug". But we can reopen discussion if needed.
Why not just eliminate the string concatenation and keep the lambda? Keeping the lambda means one less class to load from disk.