[DTrace-devel] [PATCH 2/4] test: Skip trace() of a 1-byte struct
Nick Alcock
nick.alcock at oracle.com
Tue Jul 22 13:46:06 UTC 2025
On 25 Mar 2025, eugene loh outgrape:
> From: Eugene Loh <eugene.loh at oracle.com>
>
> With commit 3a551bfd ("trace: fix char-array handling"), this test
> started to FAIL. Meanwhile, the behavior of trace() on a 1-byte
> struct is poorly defined. Users wishing clear semantics should use
> print() or other actions.
This makes trace() much, much less useful. I'd say NAK, if this means
we're going to not come up with any useful behaviour. Why not define
something, then use it?
Nobody is going to say "look at the size of something and then do a
print() rather than a trace() if it's too small". Even looking at this
commit I'm not sure what "too small" is (one byte? four bytes?
sizeof(int)? sizeof(long)? A cacheline? what?) so I don't see how our
users can be expected to.
More information about the DTrace-devel
mailing list