!AmypvmJtUjBesRrnLM:matrix.org

dataverse

191 Members
Dataverse is open source research data repository software: http://dataverse.org Public logs at https://view.matrix.org/alias/%23dataverse:matrix.org10 Servers

Load older messages


SenderMessageTime
8 Jun 2021
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Actually what would be necessary is a proper caching facade bean for the settings.18:17:51
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Removing the SettingsWrapper, cleaning up SettingsServiceBean and maybe inlining SystemConfig18:18:19
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)That will create lots of noise18:18:34
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)s/will/woud18:18:43
@oliver:bertuch.namepoikilotherm (CEST,UTC+2) * s/will/would18:18:49
@pdurbin:matrix.orgpdurbinright18:18:56
@oliver:bertuch.namepoikilotherm (CEST,UTC+2) * s/will/would/18:18:58
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Sometimes you gotta make some noise18:19:34
@pdurbin:matrix.orgpdurbinYeah. You'll get Gustavo's attention if you tell him you want to remove SettingsWrapper (and you should). That's his baby.18:20:16
@pdurbin:matrix.orgpdurbinI'm not sure the best way to sell these changes. I guess I'm thinking about what the end state might be. A bright glorious future. What does it look like? Right now people installing Dataverse can put some config values in a file. But that's not enough, right? We want fancier.18:23:14
@pdurbin:matrix.orgpdurbinI'm looking at your slides again (linked from #7639, thank you). I though we were going to refactor the email stuff as a next step.18:25:43
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Well this PR is about fixing the non-working DBSettingsSource18:26:06
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Which is already included in Dataverse today18:26:26
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Not in use, but endangering like a zombie18:26:59
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)So I thought better fix broken stuff first18:27:15
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)And no one wants to go for primary usage of MPCONFIG now18:27:47
@pdurbin:matrix.orgpdurbinI see. But the broken stuff is off the side not bothering anybody. 😀18:27:48
@pdurbin:matrix.orgpdurbin * I see. But the broken stuff is off to the side not bothering anybody. 😀18:27:55
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Lot's of people DEMAND the db settings18:28:03
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)So better not break their workflow18:28:17
@pdurbin:matrix.orgpdurbinSure, the current db settings are expected to work.18:28:54
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Oh Slava said they rely on it. They have some tooling to configure instances from some webservice18:29:22
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)(By using the DB Settings API)18:29:40
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)So the idea was to invert control. Make DB Settings overridable with MPCONFIG18:30:19
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)But SettingsWrapper is making that somewhat impossible18:30:41
@pdurbin:matrix.orgpdurbinI see.18:30:51
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)And a ViewScoped Bean doesn't seem the best fit, too. A proper caching facade, used for all of the config would make things much more inlined18:31:50
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)Today we have SystemConfig, then SettingServiceBean and SettingsWrapper specially crafted for JSF18:32:26
@pdurbin:matrix.orgpdurbinYes, we are heavily oriented toward and influenced by JSF.18:32:53
@oliver:bertuch.namepoikilotherm (CEST,UTC+2)So maybe it would be a good idea to create a unified approach. An internal settings API usable for all cases, also replacing all the System.getProperties() in the long run18:33:43

There are no newer messages yet.


Back to Room ListRoom Version: 6