| *** tristan has joined #buildstream | 06:55 | |
| *** ChanServ sets mode: +o tristan | 06:56 | |
| tristan | Happy Monday ! | 08:53 |
|---|---|---|
| tristan | I'm currently writing up a conversion script for definitions based on the .yml that ybd generates (on the unmerged ps-manifest-improvements branch) | 08:54 |
| tristan | But in the meantime, it's bright and early monday and lets get this behind us... | 08:54 |
| tristan | I copy pasted the raw html from blogs.gnome.org (and added some obligatory <p>..</p> in there): https://people.gnome.org/~tvb/reviewarea/draft.html | 08:55 |
| tristan | paulsher3ood, ... anyone... Green light finally publish ? | 08:55 |
| *** tiagogomes has joined #buildstream | 09:13 | |
| paulsher3ood | tristan: go for it | 09:44 |
| tristan | Alright ! | 09:51 |
| tiagogomes | Just a minor comment about the name of the tool. It doesn't autocomplete well on my system. | 10:08 |
| tiagogomes | I will have to type six letters before I could use autocompletion. Have you considered calling it just bstream tristan ? | 10:09 |
| paulsher3ood | +1 | 10:12 |
| paulsher3ood | let the namewars begin! | 10:12 |
| tristan | hahaha namewars | 10:16 |
| tristan | indeed I think sam suggested simply 'bst' | 10:16 |
| tiagogomes | that's a better suggestion | 10:17 |
| tristan | Ok so 2 names up for changes, I like bst because there doesnt seem to be another tool with that name, and it's nice and short | 10:17 |
| paulsher3ood | +2 for that, then | 10:18 |
| tristan | the other one is ref -> sha ... I havent done that change yet... A little skeptical because sha describes the implementation of a unique commit, but it is more accurate than 'ref' | 10:18 |
| paulsher3ood | depending on your vowel choice, it expands to 'best' or 'bust' :) | 10:19 |
| tristan | it's up on http://planet.gnome.org/ now... and post's permalink here: https://blogs.gnome.org/tvb/2017/02/06/introducing-buildstream-2/ | 10:21 |
| *** csoriano___ has joined #buildstream | 10:21 | |
| csoriano___ | Hey, glad to hear this! One question I have in mind, and I bet it's going to be the main question, is how it competes with Flatpak | 10:29 |
| tristan | csoriano___, Hi :) | 10:30 |
| csoriano___ | as in... with Flathub you will have two ways to provide your build? | 10:30 |
| csoriano___ | one with flatpak, one with buildstream | 10:30 |
| csoriano___ | ? | 10:30 |
| tristan | it's not meant to compete with flatpak really | 10:30 |
| tristan | a couple of months ago we approached Alex and discussed the big picture with him (also with the release team)... Alex asked how this would fit into the flathub picture and I offered up to provide conversion scripts (because the json is very easily convertible) | 10:32 |
| tristan | csoriano___, to be perfectly frank, I _do_ hope that after proving that BuildStream can build any flatpak, that Alex might eventually be convinced to drop or deprecate flatpak-builder (I think it's a big maintenance burden on Alex to maintain that in C) | 10:34 |
| tristan | but for now the burden of proof is on us :) | 10:34 |
| csoriano___ | oh hm, so you will build flatpaks with buildstream? | 10:34 |
| tristan | He voiced in our discussion that he will continue to support flatpak-builder on flathub for the foreseeable future | 10:34 |
| tristan | Sure | 10:35 |
| csoriano___ | but applications will still use the same flatpak json we have right now right? | 10:35 |
| tristan | That we certainly will be able to do, in anycase we will have to build runtimes and SDKs | 10:35 |
| paulsher3ood | one of the things that interests me is being able to compare the results from multiple builds, and that extends to multiple build tools that theoretically generate the same thing | 10:35 |
| csoriano___ | just that the one interpreting it will be buildstream instead of flatpak-builder | 10:35 |
| *** csoriano___ is now known as csoriano | 10:36 | |
| tristan | I would see it as a separate conversion step that can be scripted rather than a builtin loader | 10:36 |
| tristan | The main difference between how flatpak-builder and build-stream works is that we build orthogonal dependencies in parallel (that and the format) | 10:37 |
| tristan | For the demo bit that I posted, it's a manual conversion of the gedit flatpak actually | 10:38 |
| tristan | just without the metadata bits and ostree deployment to make it runnable in flatpak | 10:38 |
| csoriano | I see | 10:38 |
| csoriano | it's a pitty if you have to use even more tools to build your thing though | 10:39 |
| csoriano | I would be happy to have flatpak-builder just calling buildstream underneath | 10:39 |
| csoriano | if that's what works better | 10:39 |
| tristan | thats entirely possible, although I think there are some advantages to swapping, yaml is a bit more pleasant for one, and you can avoid more rebuilds that way | 10:40 |
| paulsher3ood | best get bst fully working first, tristan :) | 10:41 |
| tristan | i.e. with flatpak-builder everything is in sequence/linear... we can't easily introspect what really depends on what (we have to assume everything depends on the previous thing) | 10:41 |
| csoriano | I see, but it's really a pitty to have yet another tool to build your thing :( | 10:42 |
| csoriano | although I appreciate the project for things that are not with flatpak, like building gtk+ or so | 10:42 |
| csoriano | that's really useful | 10:42 |
| csoriano | and I was looking forward to that | 10:42 |
| tristan | We're trying to tie it all together indeed, I think it will be exciting to have the GNOME flatpak SDK and release modulesets in the same place | 10:43 |
| tristan | and be able to boot VMs with that | 10:43 |
| paulsher3ood | csoriano: one thing i'm hoping to see is deprecation of some existing (non-gnome) tools, if/when bst is demonstrably better than them | 10:43 |
| * tristan needs to relocate for a meeting and will reappear shortly :) | 10:43 | |
| csoriano | paulsher3ood: like? | 10:44 |
| paulsher3ood | well, the easy targets are baserock's morph and ybd tools :) | 10:45 |
| * paulsher3ood wrote ybd, fwiw | 10:45 | |
| paulsher3ood | the other thing i'm really hoping to see happen this year is bit-for-bit reproducible builds by default | 10:47 |
| *** tristan has quit IRC | 10:47 | |
| paulsher3ood | (building on the work of the debian reprobuils folks and others) | 10:47 |
| csoriano | I didn't know we were using that | 10:48 |
| paulsher3ood | gnome isn't afaik | 10:48 |
| csoriano | ah ok | 10:48 |
| csoriano | so this is work also related to non GNOME things | 10:48 |
| csoriano | well like flatpak yeah | 10:49 |
| paulsher3ood | but ostree + b4b reproducibility will be a big win | 10:49 |
| *** tristan has joined #buildstream | 11:01 | |
| *** ChanServ sets mode: +o tristan | 11:01 | |
| *** csoriano has quit IRC | 17:56 | |
| *** tiagogomes has quit IRC | 17:57 | |
| *** tristan has quit IRC | 22:06 | |
| *** jjardon has quit IRC | 23:32 | |
| *** jjardon has joined #buildstream | 23:39 | |
| *** ChanServ sets mode: +nt | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!