| *** bilelmoussaoui has quit IRC | 05:04 | |
| *** bilelmoussaoui has joined #buildstream | 05:09 | |
| *** bilelmoussaoui has quit IRC | 05:13 | |
| *** bilelmoussaoui has joined #buildstream | 05:19 | |
| *** tristan has quit IRC | 05:25 | |
| *** tristan has joined #buildstream | 05:56 | |
| *** tristan has quit IRC | 06:07 | |
| *** rdale has joined #buildstream | 07:47 | |
| *** rdale has quit IRC | 07:53 | |
| *** tristan has joined #buildstream | 08:11 | |
| *** tpollard has joined #buildstream | 08:34 | |
| *** raoul has joined #buildstream | 08:38 | |
| *** rdale has joined #buildstream | 08:50 | |
| *** laurence has quit IRC | 09:01 | |
| *** laurence has joined #buildstream | 09:01 | |
| *** jonathanmaw has joined #buildstream | 09:07 | |
| *** bilelmoussaoui has quit IRC | 09:20 | |
| *** bilelmoussaoui has joined #buildstream | 09:22 | |
| *** bilelmoussaoui has quit IRC | 09:26 | |
| benschubert | mmh our gitlab-bot seems dead again | 09:37 |
|---|---|---|
| *** bilelmoussaoui has joined #buildstream | 09:38 | |
| tpollard | being able to query a remote artifacts size++ | 09:44 |
| gitlab-br-bot | BenjaminSchubert opened MR !1368 (bschubert/fix-cython-trace->master: setup.py: Be more restrictive with BST_CYTHON_TRACE values) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1368 | 10:08 |
| gitlab-br-bot | danielsilverstone-ct approved MR !1368 (bschubert/fix-cython-trace->master: setup.py: Be more restrictive with BST_CYTHON_TRACE values) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1368 | 10:08 |
| benschubert | Is marge alive? | 10:26 |
| tpollard | nothing in the profile log for 2 days | 10:28 |
| *** bilelmoussaoui has quit IRC | 10:28 | |
| *** bilelmoussaoui has joined #buildstream | 10:28 | |
| benschubert | Could someone bump her? Or should I go to merge manually? | 10:29 |
| jennis | benschubert, I think a restart is on the way... | 10:34 |
| benschubert | thanks! | 10:37 |
| gitlab-br-bot | marge-bot123 merged MR !1368 (bschubert/fix-cython-trace->master: setup.py: Be more restrictive with BST_CYTHON_TRACE values) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1368 | 10:38 |
| benschubert | thanks! | 10:40 |
| *** bilelmoussaoui has quit IRC | 10:41 | |
| jennis | benschubert, it looks like a timeout issue, potentially this existing issue (which has just been +1'd): https://github.com/smarkets/marge-bot/issues/188 | 10:46 |
| *** pointswaves has joined #buildstream | 10:53 | |
| *** bochecha has joined #buildstream | 11:04 | |
| *** bochecha has quit IRC | 11:13 | |
| *** bochecha has joined #buildstream | 11:13 | |
| *** bochecha_ has joined #buildstream | 11:13 | |
| *** bochecha has quit IRC | 11:16 | |
| *** bochecha_ is now known as bochecha | 11:16 | |
| *** bochecha_ has joined #buildstream | 11:40 | |
| cs-shadow | so marge-bot is not on GitLab, that's a bit odd | 11:41 |
| *** bochecha has quit IRC | 11:42 | |
| *** bochecha_ is now known as bochecha | 11:42 | |
| *** tristan has quit IRC | 11:54 | |
| jennis | ha, "A merge-bot for GitLab" is the description on marge-bot's GitHub | 11:57 |
| jennis | I agree, that does seem odd | 11:57 |
| tpollard | I presume because the rest of smarkets stuff is on github | 11:58 |
| *** tristan has joined #buildstream | 12:21 | |
| WSalmon | jennis, that patch was written to fix the free desktop merge-bot, benjamb is benbrown unless i am very much mistaken. | 12:36 |
| *** lachlan has joined #buildstream | 12:43 | |
| *** jjardon has quit IRC | 12:43 | |
| *** jjardon has joined #buildstream | 12:45 | |
| *** ChanServ sets mode: +o jjardon | 12:45 | |
| *** cs-shadow has quit IRC | 12:46 | |
| *** cs-shadow has joined #buildstream | 12:46 | |
| *** benschubert has quit IRC | 12:46 | |
| *** benschubert has joined #buildstream | 12:47 | |
| *** lachlan has quit IRC | 13:01 | |
| *** phil has quit IRC | 13:01 | |
| *** phildawson has quit IRC | 13:01 | |
| *** phildawson has joined #buildstream | 13:01 | |
| *** phil has joined #buildstream | 13:01 | |
| *** mrmcq2u[m] has quit IRC | 13:16 | |
| *** mrmcq2u[m] has joined #buildstream | 13:16 | |
| *** laurence has quit IRC | 13:19 | |
| *** laurence has joined #buildstream | 13:19 | |
| *** lachlan has joined #buildstream | 13:21 | |
| benschubert | If I want to make a small breaking changes for plugins, do I need to go through the ML or do we have a simpler way to do so (I want to prevent 'node_get(type=Mapping)' to be replaced by 'node_get(type=dict)' | 13:24 |
| *** lachlan has quit IRC | 13:32 | |
| *** benschubert has quit IRC | 13:32 | |
| *** benschubert has joined #buildstream | 13:34 | |
| *** lachlan has joined #buildstream | 13:34 | |
| *** phil has quit IRC | 13:42 | |
| *** phildawson has quit IRC | 13:42 | |
| *** phil has joined #buildstream | 13:42 | |
| *** phildawson has joined #buildstream | 13:42 | |
| gitlab-br-bot | shashwatdalal opened MR !1369 (shashwatdalal/make-magic-timestamp-public->master: `utils.py`: Make `magic_timestamp` a public variable) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1369 | 13:42 |
| WSalmon | benschubert, I think changing pluging API's however trivial will effect quite a few people, including non core-developers/(people who dont interact with buildstream gitlab often) so i would think that the ML would be a good place for this. That said i would have thought that would be ok so the ML dose seem a bit much... | 14:07 |
| gitlab-br-bot | BenjaminSchubert opened MR !1370 (bschubert/optimize-node_get->master: Optimize _yaml.node_get) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1370 | 14:29 |
| *** lachlan has quit IRC | 14:33 | |
| *** lachlan has joined #buildstream | 15:11 | |
| *** lachlan has quit IRC | 15:18 | |
| *** phildawson_ has joined #buildstream | 15:24 | |
| *** phil has quit IRC | 15:25 | |
| *** phildawson has quit IRC | 15:25 | |
| *** phildawson has joined #buildstream | 15:25 | |
| *** lachlan has joined #buildstream | 15:44 | |
| *** bochecha has quit IRC | 16:02 | |
| *** rdale has quit IRC | 16:18 | |
| gitlab-br-bot | jennis opened (was WIP) MR !1344 (jennis/push_based_pipeline->master: Push based pipeline) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1344 | 16:26 |
| *** phildawson_ has quit IRC | 16:39 | |
| *** phildawson has quit IRC | 16:39 | |
| *** phildawson_ has joined #buildstream | 16:39 | |
| *** phildawson has joined #buildstream | 16:39 | |
| *** pointswaves has quit IRC | 16:46 | |
| *** ChanServ sets mode: +o tristan | 17:01 | |
| tristan | benschubert, If you happen to be looking at node_get() and the YAML changes of late, I'd invite you to at least have a glance at https://gitlab.com/BuildStream/buildstream/issues/1006 also | 17:01 |
| tristan | I think the new YAML changes are great in general as they impose more strictness and catch more corner cases, but some of it is really unwieldy | 17:02 |
| benschubert | tristan: I'll have a go in the next week(s?) to see if I can manage a better API then! Would you be happy with braking the API once again or would you rather keep a compatibility overlay? (I'd rather go with the first option, and not keep the previous cruft) | 17:04 |
| tristan | benschubert, it was already an API break in the first place, I certainly think it's fine to break again as there are no guarantees; more important to get it clean and nice in 2.x | 17:05 |
| benschubert | Ok! I'll have a look at this then | 17:06 |
| benschubert | thanks! | 17:06 |
| benschubert | WSalmon: are you around by any chance? | 17:06 |
| tristan | I think I linked in there a patch where I actually implemented a back-and-forth from dict-to-node which *might* be helpful (or might be total nonsense) | 17:06 |
| tristan | but was an easier approach for porting the debian plugins than doing all the node tedium | 17:06 |
| benschubert | found it | 17:07 |
| tristan | benschubert, fwiw, for the other question; I think the more important part here is to keep the NEWS up to date regarding breaking changes | 17:09 |
| tristan | as that is currently the only place we're documenting things, and will serve eventually as a starting point for composing a user facing porting guide | 17:09 |
| benschubert | perfect, that's what I did before :) | 17:10 |
| benschubert | tristan: for the node api refactor, I wonder if having a DictNode, a StrNode, a IntNode, a FloatNode and a ListNode would be nicer and simpler. And having each of them behave as if they were a Dict/int/float/etc ? | 17:14 |
| benschubert | Like you could do MyDictNode["blah"] = "foo" | 17:14 |
| benschubert | or have just two: One SimpleNode, and a ContainerNode, with SimpleNode.set_value(my_value) and ContainerNode.set_value(indice_or_key, value) | 17:15 |
| benschubert | so if you already had a thought about it and have a preference? Otherwise I'll decide as I'm implementing it | 17:15 |
| tristan | So... I think that being able to write python as if the nodes were dicts at least in assignment would be helpful | 17:16 |
| tristan | if you look at the referred patches where I had to do pretty wordy stuff, it mostly revolves around plugins which need to manipulate mutable nodes (for public data that is derived as a result of Element.assemble()) | 17:17 |
| tristan | The original API changes introduced seem to be pretty fine for Plugin.configure() purposes (reading configuration) | 17:17 |
| benschubert | yep, I'll make sure writing and reading the API is still nice and smoother than now | 17:18 |
| tristan | benschubert, another plausible approach might be to only have Element.[get/set]_public_data have a node <--> dict transformation ? | 17:18 |
| tristan | not sure | 17:18 |
| *** pointswaves has joined #buildstream | 17:18 | |
| tristan | I do like the idea of exposing the `Node` type directly though, rather than having to call Plugin.new_virgin_node() or whatever | 17:19 |
| benschubert | yep | 17:19 |
| benschubert | and since it's cython we can prevent access to users to things we really want to hide :) | 17:19 |
| tristan | Well, that sounds like a nice aspect of cython :) | 17:20 |
| benschubert | I don't like the node -> dict transformation for the _get_, as it means when you set it's not reflected | 17:20 |
| benschubert | on the node I mean | 17:20 |
| benschubert | or we want users to explicitely do a set_value? | 17:20 |
| tristan | I don't think so, I mean do we need strong validation at set time ? | 17:20 |
| benschubert | not sure | 17:21 |
| benschubert | I'll play more with it until I'm satisfied :) | 17:21 |
| tristan | I think it would be nice to have the node support `node['key'] = value` directly and quack like a dict in that sense | 17:21 |
| tristan | but indeed, take a look at the code which uses public data to have a feel for what's being done | 17:21 |
| benschubert | The only problem of quacking like a dict for this is that it becomes awkward for a simple node (str, int, float), as we now need node.value or something | 17:22 |
| benschubert | We could optimize for Dict/List, or have everything go through a getter/setter | 17:23 |
| tristan | I'm not sure we need our whole set of types for all basic types, we could | 17:23 |
| tristan | something to consider I guess | 17:23 |
| benschubert | I'll play with it and make a WIP MR I guess, easier to speak around code | 17:23 |
| tristan | (maybe it would mean that node members themselves would carry provenance info themselves instead of on the containing nodes ?) | 17:23 |
| tristan | it would definitely mean a lot more churn in plugins, as Plugin.node_get_member() calls would all have to be rewritten | 17:24 |
| benschubert | That would work for containers, but would not be super nice for simple types | 17:25 |
| tristan | but I mean, if we're shooting for a major improvement, that wouldnt be too bad especially if it were a one off thing | 17:25 |
| benschubert | agreed | 17:25 |
| benschubert | I'd rather have a really nice API for bst 2.0 :) | 17:25 |
| tristan | Then at the same time we'd feed a strongly (or more 'specifically') types Node to Plugin.configure(), and we'd have plugins call Node.get_member(), and just nuke Plugin.node_get_member() completely | 17:26 |
| tristan | nuke all the Plugin.node_*() funcs | 17:26 |
| tristan | s/types/typed | 17:26 |
| *** raoul has quit IRC | 17:34 | |
| *** pointswaves has quit IRC | 17:37 | |
| Kinnison | tristan: all nodes carry their provenance now | 17:41 |
| Kinnison | That was part of the new world order | 17:41 |
| tristan | Kinnison, except that from what I understood, and I could be wrong... nodes are still just the nodes, not the values | 17:43 |
| Kinnison | Nope values are nodes too internally | 17:43 |
| tristan | Kinnison, i.e. is the key/value pair within a node... also a node ? | 17:43 |
| tristan | I see | 17:43 |
| Kinnison | Though we currently simplify out from there at the api boundary | 17:43 |
| tristan | I didn't understand that from the API | 17:43 |
| tristan | right | 17:44 |
| benschubert | tristan: yup that seems nice. | 17:44 |
| benschubert | tristan, kinnison, what about Node.get_value(type=my_type), Node.get_from_list(indice, type=my_type) and Node.get_from_dict(key, type=my_type) ? | 17:44 |
| Kinnison | So there's a good chance that your idea of reworking Node to have member methods etc might fly | 17:44 |
| Kinnison | I think perhaps Node.as_foo() maybe? | 17:45 |
| Kinnison | And indexing by number treats as list, indexing by stri g as dict? | 17:45 |
| Kinnison | Bad phone keyboard. Apologies. | 17:45 |
| benschubert | mmh that would work yep | 17:45 |
| Kinnison | Perhaps Node.assert_footype() for checks? | 17:46 |
| Kinnison | I'm guessing at an api right now, | 17:46 |
| Kinnison | Will revisit this tomorrow. I need to go make dinner. | 17:47 |
| benschubert | Or just raise when the type is not good | 17:49 |
| *** lachlan has quit IRC | 18:13 | |
| *** lachlan has joined #buildstream | 18:16 | |
| *** lachlan has quit IRC | 18:19 | |
| *** lachlan has joined #buildstream | 18:21 | |
| *** lachlan has quit IRC | 18:36 | |
| *** xjuan has joined #buildstream | 18:42 | |
| *** lachlan has joined #buildstream | 18:53 | |
| *** lachlan has quit IRC | 18:59 | |
| *** lachlan has joined #buildstream | 19:07 | |
| *** lachlan has quit IRC | 19:10 | |
| *** lachlan has joined #buildstream | 19:13 | |
| *** xjuan has quit IRC | 19:26 | |
| *** lachlan has quit IRC | 19:48 | |
| *** brlogger has joined #buildstream | 20:07 | |
| *** slaf has quit IRC | 20:07 | |
| *** verdre[m]1 has quit IRC | 20:07 | |
| *** segfault3[m] has quit IRC | 20:07 | |
| *** tlater[m] has quit IRC | 20:07 | |
| *** ssssam[m] has quit IRC | 20:07 | |
| *** skullone[m] has quit IRC | 20:07 | |
| *** asingh_[m] has quit IRC | 20:07 | |
| *** jjardon[m] has quit IRC | 20:07 | |
| *** ironfoot has quit IRC | 20:07 | |
| *** brlogger` has quit IRC | 20:07 | |
| *** admin has quit IRC | 20:07 | |
| *** WSalmon has quit IRC | 20:07 | |
| *** valentind has quit IRC | 20:07 | |
| *** cgmcintyre[m] has quit IRC | 20:07 | |
| *** kailueke[m] has quit IRC | 20:07 | |
| *** albfan[m] has quit IRC | 20:07 | |
| *** pro[m] has quit IRC | 20:07 | |
| *** dineshdb[m] has quit IRC | 20:07 | |
| *** Trevinho[m] has quit IRC | 20:07 | |
| *** awacheux[m] has quit IRC | 20:07 | |
| *** Demos[m] has quit IRC | 20:07 | |
| *** connorshea[m] has quit IRC | 20:07 | |
| *** reuben640[m] has quit IRC | 20:07 | |
| *** theawless[m] has quit IRC | 20:07 | |
| *** nielsdg has quit IRC | 20:07 | |
| *** dbuch has quit IRC | 20:07 | |
| *** doras[m] has quit IRC | 20:07 | |
| *** waltervargas[m] has quit IRC | 20:07 | |
| *** coldtom has quit IRC | 20:07 | |
| *** brlogger has joined #buildstream | 20:07 | |
| *** ironfoot has joined #buildstream | 20:09 | |
| *** coldtom has joined #buildstream | 20:10 | |
| *** lachlan has joined #buildstream | 20:12 | |
| *** swick has joined #buildstream | 20:12 | |
| *** slaf has joined #buildstream | 20:13 | |
| *** valentind has joined #buildstream | 20:13 | |
| *** lachlan has quit IRC | 20:15 | |
| *** jonathanmaw has quit IRC | 20:16 | |
| *** WSalmon has joined #buildstream | 20:21 | |
| *** dineshdb[m] has joined #buildstream | 20:28 | |
| *** Demos[m] has joined #buildstream | 20:29 | |
| *** nielsdg has joined #buildstream | 20:38 | |
| *** kailueke[m] has joined #buildstream | 20:54 | |
| *** segfault3[m] has joined #buildstream | 20:56 | |
| Kinnison | perhaps | 21:08 |
| *** waltervargas[m] has joined #buildstream | 21:11 | |
| *** ssssam[m] has joined #buildstream | 21:19 | |
| *** awacheux[m] has joined #buildstream | 22:01 | |
| *** pointswaves has joined #buildstream | 22:05 | |
| *** doras[m] has joined #buildstream | 22:06 | |
| *** reuben640[m] has joined #buildstream | 22:17 | |
| *** skullone[m] has joined #buildstream | 22:38 | |
| *** cgmcintyre[m] has joined #buildstream | 22:52 | |
| *** connorshea[m] has joined #buildstream | 23:06 | |
| *** tlater[m] has joined #buildstream | 23:28 | |
| *** jjardon[m] has joined #buildstream | 23:28 | |
| *** theawless[m] has joined #buildstream | 23:42 | |
| *** pointswaves has quit IRC | 23:44 | |
| *** dbuch has joined #buildstream | 23:47 | |
| *** albfan[m] has joined #buildstream | 23:51 | |
| *** Trevinho[m] has joined #buildstream | 23:54 | |
| *** asingh_[m] has joined #buildstream | 23:56 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!