1. 43
  1. 6

    This seems like a task that could be amenable to evolutionary algorithms but there’s so few projects of this scope there’s very few groups that have the motivation to do it. Props to Ingo for putting in so much manual work and persevering through the first few disappointing deadends.

    1. 2

      In Games we have various brute force approaches to this, trying to figure out the best options of forward including and reducing the actual number of includes.

      With this set of changes is there anything touched beyond #include lines and forward declarations? Like did they do things like move individual structs out to their own files so that they can IWYU?

      1. 2

        With this set of changes is there anything touched beyond #include lines and forward declarations?

        Yes, there are substantial changes to introduce per_task machinery modeled after existing per_cpu, which allows you to extend task_struct without depending on its definition.

    2. 6

      More of a side comment than a comment on the amazing work going on here, which should result in many companies seeing productivity gains, and probably even reductions in energy usage around the world… but are LWN comments always this terrible?

      oH My GoD, SoMeOnE UsEd An EMOJI!!!

      I can’t wait to use this! Well you’ll have to!!!

      With basically no discussion about the actual content of this change. I expected much better from the Linux community.

      1. 6

        There are some nuggets of knowledge to find there, but yeah, overall quality is low. Basically it’s a void people vent into, same as the comment section on any popular news website, or comments under a FB repost of some news. It’s not a discussion really, people in that situation rarely go back and engage on a deeper level.

        (I like to imagine any comment section under a news article as a bar full of half-drunk people shouting at the report on the news channel)

        1. 1

          By that logic, shouldn’t the same apply to this site?

          1. 8

            IMO not for two reasons:

            • this site exists for the purpose of the discussion rather than having comments available on a side and bad-quality stuff gets moderated away fairly quickly
            • the lower popularity translates to less drunkenness in the analogy ;)

            The orange site is definitely playing with the threshold on the second point.

            1. 1

              How did you know I originally wanted to say “here and on the orange site” then thought better of it!? ;-)

            2. 3

              Moderation and invite-only together with a non “me too” and no politics policy helps a lot, I would like to think. Also the hotness modifier for rants and removal of pitchfork posts.

          2. 2

            This LWN article was linked from many places, so low quality commenters flooded in. I think that’s all there is to it in this case.

            1. 1

              Are LWN comments a good indicator for the whole “Linux community” (no, probably not…)

              1. 11

                Comment sections are always indicative of some parts of a community. It’s almost always the worst part.

                1. 3

                  It’s almost always the worst part

                  Though it’s probably similar to local newsletters about IT and technology here: A lot of people with dangerous semi knowledge being noisy, which aren’t actually using the technology or aren’t using it as a developer and only as the end user (which doesn’t care at all how fast the linux kernel is compiled when installed via apt).

                  1. 2

                    It would have been polite to put “present company excluded” there, to make people feel good even if it’s untrue.

                    1. 2

                      I think that Lobste.rs is much more intentional a community than a comment section. The relationships between the folks and the culture around those relationships is more important to maintaining a community than anything else; a lot of the strictures that some people hypothesize as being causal are probably only correlates of that culture. But culture is very hard to build intentionally, and there are only a few places where it’s worked (I think of Metafilter, for instance).