LWN: Comments on "Linux ARM mailing lists moving"

Linux ARM mailing lists moving

eparis123 — Mon, 24 Aug 2009 23:47:55 +0000

I understand such an argument, and possibly such an environment gives a fuzzy sense of a more-connected group than an open-list.

Unfortunately such a method can't scale well with a project as big as the kernel as evidenced by Russel just burning out. Also, some people may not be interested in subscribing to a certain list and just desire posting a small change here or there (say a small context-insensitive internal API change).

Lots of sensible things become insensible when given in a different context ;)

Linux ARM mailing lists moving

ibisum — Mon, 24 Aug 2009 14:01:58 +0000

I think the argument is that you shouldn't be submitting patches if you're not already subscribed to the list, the point being that the list is the primary means of coordination between patch-contributors. It certainly is an administrative nightmare to receive patches from individuals who are not aware of each others' work, so it would be sensible to have all patches come through the subscribers-only list.

Well, of course, sensible. But not popular. Popular is not always sensible.

Linux ARM mailing lists moving

eparis123 — Mon, 24 Aug 2009 12:39:00 +0000

Those subscribers-only lists are really irritating when sending a big patchset.

Assuming sending 12 patches in a series, and say two of them was a related to subsystem X, so those two was CCed to X mailing list but then you discover they get bounced.

You have to _interrupt_ your workflow, subscribe, wait approval, and _then_ send those two again. This totally breaks the neat threading layout of the patch series and leads to synchronization problems in patches review (conflicting suggestions at times between reviewers).

Thanks to Russel for his previous work, but he has to adapt the patches and reviewing system, not the system adapting his.