Re: coccinelle: merge two rules from flex_alloc.cocci
- From
SZEDER Gábor <szeder.dev@gmail.com>
- Date
- Nov 14, 2019, 16:35 UTC
- Message-ID
- <20191114163527.GT4348@szeder.dev>
- In-Reply-To
- <1d08b49e-1f41-4290-a64b-dad9fd2288de@web.de>
On Thu, Nov 14, 2019 at 09:15:47AM +0100, Markus Elfring wrote:
> > If you could have some before/after numbers, that would be cool. > > Does any test infrastructure (or benchmarks) exist which you would trust for > corresponding comparisons of software run time characteristics?
Yes, just run:
make cocciclean time make contrib/coccinelle/flex_alloc.cocci.patch
before and after your changes, and include the timing results in the commit message if there is a notable difference. If it gets faster, great! If it gets slower, then update the commit message with a convincing argument about why the change is worth the performance penalty.
FWIW, I did just that with your "coccinelle: merge twelve rules from object_id.cocci" patch [1], and the runtime went down from 2m48.610 to 2m34.395, a bit over 8% speedup. (with Ubuntu 16.04's Coccinelle 1.0.4)
[1] https://public-inbox.org/git/6c9962c0-67c1-e700-c145-793ce6498099@web.de/