Hello. 2020-06-06 13:47 UTC+02:00, Xeno Amess <xenoam...@gmail.com>: > Yes you are right. >>you must squash the commits. >>the commit messages are uninformative > > Usually I'd do them when commiter think this pr is ok,
As said before, it's needlessly difficult to figure out the net changes when they result from several commits. > and then tell me we > should merge it, and at that moment I squash it, and it get merged. > the commit messages will be fixed at that time too. The commit message should not need any fixing after the commit. It must correspond to what was done, and would often help the reviewer. [Why make it more complicated for us to accept the PR?] >>partly revert a previous change > > Yes, as my newest performance test shows something is not quite right in > my previous changes. It makes the codes even slower. > So I changed that part. But this back-and-forth is not something interesting to keep in the history if we can avoid it. > >>benchmark are indeed necessary > > agreed, I did a benchmark yesterday, and pointed a link on github pages. > I'll copy/paste > >>The PR contains different types of changes: one is "use arraycopy" but the > others are not documented anywhere (commit message, JIRA). > Ah, I forgot to create a jira ticket for it. sorry about that. > > So, in conclusion, I will close this pr, and make two jira tickets, for the > two different types of refines. Thanks, Gilles > > [...] --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
