From 156e721cb97e277ef3786af458d4e6ddf3a5387d Mon Sep 17 00:00:00 2001 From: Daniel Silverstone Date: Sat, 5 Dec 2020 15:19:14 +0000 Subject: Meeting notes from 5th Dec Signed-off-by: Daniel Silverstone --- blog/2020/12/05/iteration.mdwn | 152 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 152 insertions(+) create mode 100644 blog/2020/12/05/iteration.mdwn diff --git a/blog/2020/12/05/iteration.mdwn b/blog/2020/12/05/iteration.mdwn new file mode 100644 index 0000000..101d40e --- /dev/null +++ b/blog/2020/12/05/iteration.mdwn @@ -0,0 +1,152 @@ +[[!meta title="Iteration planning meeting: December 5th"]] +[[!meta date="2020-12-05 13:30"]] +[[!tag meeting]] + +# Actions from last time + +- Daniel to file an issue about making codegen require an explicit + template declaration. -- [[!issue 128]] was filed (and fixed) +- Daniel to review open merge requests from Lars. -- Done +- Lars and Daniel to schedule a discussion about running scenarios + conditionally. -- Done, Lars updated [[!issue 20]] with results of + conversation. + +# This meeting + +We finished [[!milestone 18]], having fixed [[!issue 124]], [[!issue 102]], +[[!issue 122]], and [[!issue 107]]. + +Progress was made on [[!issue 123]], [[!issue 101]], [[!issue 20]]. + +We agreed this iteration was probably too large over all, and this makes us +sad because we can't finish everything. We were all pleased that Dan got to +contribute again. + +We closed [[!milestone 18]] and created [[!milestone 19]]. + +# Pending MRs + +There were no pending merge requests in subplot-web. + +There is a WIP for rust codegen, and a daemon library MR. + +## Daemon library + +Daniel noted that Lars had been a very naughty boy and use `sleep()`. This +approach is bad because it can waste time and it also can be insufficient for +its intended purpose. Lars marked this as draft in order to not have it merged +yet. + +## Rust codegen + +Lars has skimmed over it. It uses parts of Rust which Lars isn't familiar with. +It doesn't work with Buster's rustc which is a problem. It turns out the `matches!()` +macro is a problem on Buster's rustc. Lars is willing to drop the requirement +that we work with Buster's rustc at this point. + +# Branch review + +Subplot-web had a branch from Lars' trialling which he has now cleaned up. +Subplot itself only has branches for the open MRs. + +# Issue review + +Subplot-web has no issues. + +Subplot itself has 30 open issues, last time we went over them all, so this +meeting we only looked at changed / new issues. + +- [[!issue 127]] - This was an issue Lars filed while working on something using + subplot. We labelled this codegen, feature, helpwanted, somedaymaybe. +- [[!issue 129]] - Regarding architectures other than amd64 - Lars' coworker + reported that whenever he runs `apt update` he gets a warning that the subplot + repository does not support i386. + Daniel added a note about adding `[arch=amd64]` instead. Lars intends to + alter the script to do this. Lars retitled the issue and tentatively assigned + it to the new iteration. +- [[!issue 130]] - Came up while Lars was pairing with Dan. Lars said he could + do this pretty easily, so assigned it to himself and into the new iteration. + Daniel pointed out it's important that our CI uses the script if we write it. +- [[!issue 131]] - Daniel labelled this someday-maybe actively-discussed hard + feature. Daniel also tentatively put it into the iteration for discussion. +- [[!issue 101]] - Daniel moved it to the new iteration. +- [[!issue 20]] - We discussed Daniel's comment about cleanups, and after some + careful pondering we came to an agreement which Lars documented on the issue. + Daniel added a bullet pointed action plan to the top of the issue, removed the + actively-discussed label, and created and added a tracking label. + Daniel marked this as blocked by [[!issue 132]] which was tentatively put into + this new iteration +- [[!issue 123]] - This is going into the new iteration as Lars is working on + this already and has a WIP MR. + +## Issues in the old iteration which were not completed + +- [[!issue 120]] - Untargetted for now +- [[!issue 117]] - Tentatively added to the new iteration +- [[!issue 116]] - Likewise. +- [[!issue 108]] - Likewise. +- [[!issue 96]] - Untargetting for now +- [[!issue 74]] - Likewise. +- [[!issue 6]] - Likewise. + +# Planning the next iteration + +We then went through our tentative plan for 9 issues in [[!milestone 19]]. + +We assigned [[!issue 131]] to Daniel and Lars. + +# Discussing the WIP Rust MR. + +Daniel raised the point that the typemap in the MR is just string to string +currently. Lars thinks it's a good idea to have this and to have it checked +in the AST pass. + +In terms of extraction of the feature, we should use the typemap to help guide +the construction of regular expressions from simple patterns, instead of the +current optional typing in the simple patterns. + +Daniel described how `Option` will work eventually. + +Daniel mentioned the new filters he added to Tera, Lars asked for it to be +split into a separate MR. + +Lars intends to file an MR to add `.tera` to the other template names. + +# Regarding the tracking of getting new users to Subplot + +We added [[!issue 116]] as a blocker to [[!issue 118]], labelling [[!issue 125]] +as a tracking issue. + +Lars is concerned that we're still not getting people to try Subplot. Perhaps +we could redo the Fable experiment. We will plan for the next iteration (i.e. +the one starting Dec 20th) to include efforts toward this. + +# Regarding Subplot in Debian + +Daniel marked [[!issue 119]] as a tracking issue and marked it blocked by [[!issue 125]]. + +We're not in a position to support subplot for 2 years yet. As such we're not +aiming it at Bullseye + +# Divergence between codegen templates + +Python template has `--env` the Bash one does not. On the one hand, we don't +necessarily want to support CLI options in the generated bash. On the other +hand we feel bad if bash is a third class citizen. + +Daniel suggested that all runners are going to have their own individual sets of +properties/capabilities. As such, rather than worrying about making all the +runners have the same capabilities, we should have a table of what has what. + +# Actions + +- Daniel to extract `cargo fmt` commit from [[!mr 105]] into a new MR for merging. +- Daniel and Lars to set up a call about [[!issue 131]]. +- Daniel to split typemap work out into a separate MR so that it can be reviewed + and merged independently of the Rust runner work. (Daniel to file an issue + before doing this to include our decided approach). +- Lars to file MR to add `.tera` to bash and python template +- Daniel to file an MR for splitting out the filters in the Rust MR. +- Daniel to file an issue about embedding templates with a view to `cargo install` + related to [[!issue 66]] and [[!issue 109]]. +- Lars to open an issue about how we don't have the feature table for runners. -- cgit v1.2.1