On 9 November 2015 at 17:40, Dr. David Alan Gilbert <dgilb...@redhat.com> wrote: > * Peter Maydell (peter.mayd...@linaro.org) wrote: >> 57 patches three days before hard freeze? Are you *really* sure about this? > > Well: > a) It's been posted and tested for a long time now; most things seem happy. > b) It's been posted and reviewed multiple times, the changes since the > previous version are purely cleanups from previous review and one > fix. > c) We've been feeding some smaller parts in, individually, but a lot > of the changes are only used by later parts of the code base, and > people don't like changes going in that aren't used yet. > d) I'm here for those 3 days and after to fix it if it explodes.
I'm putting it through the merge process, but I really don't like huge pulls hitting the tree very late in softfreeze. The idea of softfreeze is supposed to be that we start out with a lot of churn and gently drop down towards "no changes except bugfixes" as we hit the hardfreeze date. Big changesets are more likely to be destabilizing, and destabilizing changes are worse the later in the freeze process they hit. Also, by dropping a big pullrequest late you are significantly increasing your risk that compile failures and other "only noticed in the merge process" problems will take long enough to fix that you fail to get the thing in before the hardfreeze deadline. Any pull request of 50 patches or more can end up running into several of that kind of problem and need several tries to get through. (For instance, you've got a windows build failure, which I'm about to send a separate mail for.) thanks -- PMM