Logs: freenode/#haskell
| 2021-04-08 18:35:55 | → | jjhoo joins (jahakala@dsl-trebng21-b048b5-171.dhcp.inet.fi) |
| 2021-04-08 18:36:00 | <sclv> | the creds i care less about tbqh |
| 2021-04-08 18:36:00 | → | royal_screwup21 joins (52254809@gateway/web/cgi-irc/kiwiirc.com/ip.82.37.72.9) |
| 2021-04-08 18:36:15 | → | hendursaga joins (~weechat@gateway/tor-sasl/hendursaga) |
| 2021-04-08 18:37:46 | <sclv> | btw sm[m] git-annex doesn't let you "import that elsewhere" nicely afaik? it would be great if it did! |
| 2021-04-08 18:37:55 | <sclv> | it just ensures you have the data, which is in itself worthwhile |
| 2021-04-08 18:39:06 | <sm[m]> | sclv: yes, actually recreating it as a working tools would be a big job. Not yet easy at all |
| 2021-04-08 18:39:43 | <sm[m]> | hopefully before too long this kind of data will have a standard schema and multiple apps to load it into |
| 2021-04-08 18:40:08 | <sclv> | i think that'll require more competition than i see existing anytime soon |
| 2021-04-08 18:40:13 | <sclv> | laziness and lock-in are strong forces |
| 2021-04-08 18:40:16 | <sm[m]> | though I guess that probably won;t happen until github screws up.. right |
| 2021-04-08 18:40:38 | <sclv> | right, lol was going to say the same thing. M$ giving people a scare would really inject some urgency |
| 2021-04-08 18:40:42 | → | ezzieyguywuf joins (~Unknown@unaffiliated/ezzieyguywuf) |
| 2021-04-08 18:40:47 | <maerwald> | github is just a source code mirror |
| 2021-04-08 18:41:37 | <maerwald> | real haskellers use darcsden :p |
| 2021-04-08 18:42:11 | <sm[m]> | darcs hub! darcsden is old hat :) |
| 2021-04-08 18:43:40 | <sm[m]> | darcsweb.cgi -> patch-tag.com -> darcsden.com -> hub.darcs.net |
| 2021-04-08 18:43:48 | × | m0rphism quits (~m0rphism@HSI-KBW-085-216-104-059.hsi.kabelbw.de) (Quit: WeeChat 2.9) |
| 2021-04-08 18:43:55 | <maerwald> | I think the two main reasons for the success of github etc are: 1. free CI and 2. UI usability. Beating the UI usability is really hard |
| 2021-04-08 18:44:21 | <sm[m]> | the CI is great, but just a recent addition |
| 2021-04-08 18:44:34 | <sm[m]> | UI usability and network effect I'd say |
| 2021-04-08 18:45:21 | <sclv> | out of the box, git leaves a lot to be desired. you really need some gui chrome to make it a nice workflow |
| 2021-04-08 18:46:11 | → | aerona joins (~aerona@2600:6c54:4600:f300:ec35:c665:ee7f:e84a) |
| 2021-04-08 18:46:16 | × | royal_screwup21 quits (52254809@gateway/web/cgi-irc/kiwiirc.com/ip.82.37.72.9) (Quit: Connection closed) |
| 2021-04-08 18:46:31 | <sm[m]> | agreed. It first became usable/useful to me when I gathered a few decent command aliases in .gitconfig (then magit, IDEs etc.) |
| 2021-04-08 18:47:00 | → | royal_screwup21 joins (52254809@gateway/web/cgi-irc/kiwiirc.com/ip.82.37.72.9) |
| 2021-04-08 18:47:35 | <maerwald> | I only use tig |
| 2021-04-08 18:48:11 | <sclv> | i think its as simple as the fact that the whole "PR" concept is fundamental to modern workflows, and doesn't exist natively in git proper at all |
| 2021-04-08 18:48:56 | <sm[m]> | some say PRs are bad. I forget what the alternative is |
| 2021-04-08 18:49:10 | <maerwald> | sm[m]: patches |
| 2021-04-08 18:49:22 | × | ixlun quits (~matthew@213.205.241.25) (Read error: Connection reset by peer) |
| 2021-04-08 18:49:26 | <maerwald> | PRs are pretty native |
| 2021-04-08 18:49:31 | <sm[m]> | isn't a PR a patch plus notification ? |
| 2021-04-08 18:49:58 | <geekosaur> | it's a patch implemented as a branch |
| 2021-04-08 18:50:00 | <maerwald> | a PR can simple be an email (as was done by the linux workflow) giving you a link to a repo, info of branch name |
| 2021-04-08 18:50:10 | <sm[m]> | I think we don't like github PRs because of all the ceremony involved - fork, branch, etc. |
| 2021-04-08 18:50:10 | <maerwald> | no, PRs are not patches |
| 2021-04-08 18:50:33 | <yushyin> | it even has a manpage (man git-request-pull) |
| 2021-04-08 18:50:47 | <maerwald> | branches are not patches either |
| 2021-04-08 18:50:50 | <sclv> | PRs being integrated with / coupled to discussion and intertwingled with issue tracking feels like it was a github innovation |
| 2021-04-08 18:50:52 | <maerwald> | commits are not patches |
| 2021-04-08 18:51:06 | <hyperisco> | patches are not patches either, if you really scrutinize it |
| 2021-04-08 18:51:33 | <maerwald> | patch workflow can be done with git too (git send-email) |
| 2021-04-08 18:51:38 | <maerwald> | but a PR is not that |
| 2021-04-08 18:51:43 | <sclv> | no i've thought about it a lot and a patch is a patch |
| 2021-04-08 18:51:44 | × | royal_screwup21 quits (52254809@gateway/web/cgi-irc/kiwiirc.com/ip.82.37.72.9) (Ping timeout: 268 seconds) |
| 2021-04-08 18:52:04 | × | DavidEichmann quits (~david@47.27.93.209.dyn.plus.net) (Remote host closed the connection) |
| 2021-04-08 18:52:06 | × | zebrag quits (~inkbottle@aaubervilliers-654-1-2-51.w83-200.abo.wanadoo.fr) (Read error: Connection reset by peer) |
| 2021-04-08 18:52:27 | → | zebrag joins (~inkbottle@aaubervilliers-654-1-2-51.w83-200.abo.wanadoo.fr) |
| 2021-04-08 18:52:38 | <hyperisco> | I'm glad someone has looked into it for us |
| 2021-04-08 18:52:53 | <davean> | sm[m]: git pull rrequests and github bull requests are somewhat different is one of the complaints |
| 2021-04-08 18:53:01 | <maerwald> | and the advantage of PR workflow is, that you don't have to deal with patches |
| 2021-04-08 18:53:10 | <maerwald> | you can just checkout out the branch and test |
| 2021-04-08 18:53:14 | <maerwald> | with no applying of patches |
| 2021-04-08 18:53:29 | <davean> | Right which was a git thing, but the way github does it mixed it with other things |
| 2021-04-08 18:53:44 | <hyperisco> | "feature branch" |
| 2021-04-08 18:54:06 | <davean> | no, it was litterly "git pull", not feature branches |
| 2021-04-08 18:54:12 | <davean> | because git is desentralized |
| 2021-04-08 18:54:38 | <davean> | see kernel.org |
| 2021-04-08 18:55:07 | <hyperisco> | I have never worked with source control in a decentralised way |
| 2021-04-08 18:55:19 | <davean> | I have, I much prefer it |
| 2021-04-08 18:55:20 | × | hidedagger quits (~nate@unaffiliated/hidedagger) (Quit: WeeChat 3.1) |
| 2021-04-08 18:55:37 | <maerwald> | it can be quite annoying wrt CI :) |
| 2021-04-08 18:55:46 | <hyperisco> | makes sense for open source where everyone needs their own variants |
| 2021-04-08 18:55:51 | <davean> | Ci is run differently for it |
| 2021-04-08 18:56:09 | <davean> | if you try to fit CI as you expect for a centralized system on a distributed one, of course theres issues |
| 2021-04-08 18:56:17 | <davean> | CI just looks entirely different |
| 2021-04-08 18:56:32 | <davean> | I'll point out the commit IDs are still globally stable when distributed |
| 2021-04-08 18:56:48 | <davean> | So its not even hard to do the CI for it |
| 2021-04-08 18:57:06 | <maerwald> | e.g. my GHC PR failed, because I don't have permissions for some of the gitlab runners or something |
| 2021-04-08 18:57:42 | <maerwald> | s/GHC/ci-images/ |
| 2021-04-08 18:57:53 | <davean> | But you're sticking a centralized approach on a distirbuted solution to get that issue there |
| 2021-04-08 18:58:09 | <davean> | Thats exactly why I said you couldn't bring that centralized mindset and get good results |
| 2021-04-08 18:58:15 | × | frozenErebus quits (~frozenEre@37.231.244.249) (Ping timeout: 260 seconds) |
| 2021-04-08 18:58:17 | <maerwald> | I've never seen a decentralized CI workflow that works |
| 2021-04-08 18:58:27 | <sm[m]> | what's the essence of "source control in a decentralised way", compared to github, davean ? More repos, hosted in more places ? |
| 2021-04-08 18:58:31 | × | heatsink quits (~heatsink@108-201-191-115.lightspeed.sntcca.sbcglobal.net) (Remote host closed the connection) |
| 2021-04-08 18:58:40 | <davean> | sm[m]: Everyone has their own seperate repos |
| 2021-04-08 18:58:50 | <davean> | everything is pull based, not push based |
| 2021-04-08 18:59:13 | <maerwald> | yeah, github doesn't have separate repos internally if you fork, so that was a common attack vector for various jokers |
| 2021-04-08 18:59:31 | <maerwald> | so if you fork, make sure to not fork :p |
| 2021-04-08 18:59:58 | <sclv> | there's a huge irony in that git was designed to be distributed, and the whole appeal of github is it recentralized everything |
| 2021-04-08 19:00:34 | <davean> | sclv: Hum, thats why I don't use github isn't it? ;) |
| 2021-04-08 19:01:03 | <sclv> | imho decenralized discussion is not a fully solved problem, i'll note |
| 2021-04-08 19:01:18 | <yushyin> | ^ this |
| 2021-04-08 19:01:22 | <maerwald> | sclv: yes, it doesn't work |
| 2021-04-08 19:01:40 | <sm[m]> | davean: on github, everyone making pull requests has their own repo, and it's pull based, too |
| 2021-04-08 19:01:42 | <maerwald> | that's why we have this mess of ecosystem migrations in haskell |
| 2021-04-08 19:01:49 | <sm[m]> | at least, IME |
| 2021-04-08 19:02:37 | <maerwald> | some ppl use github, some not, some only respond via emails |
| 2021-04-08 19:02:58 | <maerwald> | it's impossible to coordinate anything |
| 2021-04-08 19:03:06 | <davean> | sm[m]: no, it very much isn't |
| 2021-04-08 19:03:13 | → | minoru_shiraeesh joins (~shiraeesh@5.101.59.91) |
| 2021-04-08 19:03:36 | <maerwald> | sm[m]: internally it's one repo if you fork |
| 2021-04-08 19:03:42 | <davean> | Its not even interally |
| 2021-04-08 19:03:46 | <maerwald> | the UI makes you think it's separate |
| 2021-04-08 19:03:47 | <davean> | that shows up VERY clearly |
| 2021-04-08 19:03:47 | × | __minoru__shirae quits (~shiraeesh@109.166.57.223) (Ping timeout: 248 seconds) |
| 2021-04-08 19:03:53 | <davean> | And its very much push based |
| 2021-04-08 19:04:09 | <sm[m]> | what do you mean by push based here ? |
All times are in UTC.