On Thu, Oct 06, 2022 at 10:39:22AM +0200, Krzysztof Kozlowski wrote: > On 06/10/2022 10:21, Johan Hovold wrote: > >> What to track: > >> 1. Which bindings to convert to YAML, > >> 2. Missing compatibles (either entirely or because of missing conversion), > >> 3. `dt_binding_check` warnings (usually connected with 1-2), > >> 4. `dtbs_check` warnings. > >> > >> Rob's bot gives us daily output for 1-4, but how can we track current > >> efforts to avoid duplication of work? Also it would allow people to find > >> tasks for them to get contributions to Linux kernel :). Is anyone in > >> community interested in tracking it together, in a public way? > > > > Is this a real problem that needs fixing? I mean how often does it > > happen that people submit the same YAML conversion for example? Since it > > doesn't take that long to do a conversion, I'm not sure what tracking > > this on some webpage buys us. It's better to just search lore before > > starting a new conversion. Or search the linux-next tree to see what's > > still pending. > > In terms of DT bindings conversion to DT schema: > If I were not checking for ongoing work, I would duplicate effort like > ~10 times. Few other folks hit it few times, at least. Several bindings > are being converted for ~1 year! Ok, but the conversion itself doesn't take that long even if getting it merged and fixing up new warnings may take some time. And after the initial posting, a quick lore search allows you to find any on-going conversion efforts. Perhaps that can just be mentioned in a wiki-page of sorts that lists remaining conversions with some suggestions for how best to go about things. > In terms of DTS warnings - it's difficult even to check/search. For what > do you search? Warnings? Pretty often they are not part of commit msg. > By file? Then you might have many, many unrelated search results. I wasn't suggesting to use lore for warnings, but searching the lists for changes to a particular dts before embarking on a clean up doesn't seem unreasonable. > > Similarly for the other points above, as it doesn't take very long to > > add a missing compatible or fix a warning it seems a bit excessive to > > try to track this manually. > > True, some are trivial. Some however need fixing the binding which takes > time. Right. > > Perhaps a list of pending conversions or missing compatibles could be > > useful for someone who's short on work, but it's bound to get outdated > > pretty quickly. > > Another point is to have the visibility on the amount of work to be > done. But I understand that's maybe topic just for few, e.g. me, so I > can just track stuff for myself. Yeah, I don't think that any such extra process should be needed outside a small group that may potentially be working on cleaning up bindings and dts in bulk. If you were to maintain such a list of pending and on-going conversions for yourself, perhaps making that public is all that's needed here? Johan
