biff

jf 2024-09-27T02:12:17.729929Z

if I'm thinking of changing a biff application to XTDB 2, how much effort would this be / what needs to change?

seancorfield 2024-09-27T02:19:48.612719Z

https://github.com/jacobobryant/biff/tree/xtdb2 -- I don't know how far @foo got with that tho'...

seancorfield 2024-09-27T02:21:19.752829Z

...if I remember his posts elsewhere, one of the big questions remaining was what deployment looks like for the solo developer...?

2024-09-27T02:32:34.167419Z

yep--see the accompanying blog post also: https://biffweb.com/p/trying-out-xtdb2/. for deployment see this post: https://discuss.xtdb.com/t/deployment-recommendations-for-small-apps/483 (summary is you can either go with the filesystem storage or upgrade to redpanda, depending on how much of a Serious Project you're building)

jf 2024-09-27T02:33:06.828409Z

ok. Thank you both! I will have a look and a think šŸ™‚

seancorfield 2024-09-27T02:55:09.280629Z

I think Biff was fairly tightly fitted to XT1 so some of the "lack of fit" for XT2 is due to that -- and the big differences between XT1 (more of a Datomic competitor, IMO) and XT2 (a bitemporal document-based database with full SQL support). And for XT2, SQL is sort of the preferred API so it's an easier fit for web apps that are more traditional in that respect. When I initially created a version of usermanager for XT2, I relied on my experimental next.jdbc / XTQL bridge, but I recently converted it to use the pure SQL approach (and plain next.jdbc with the PostgreSQL JDBC driver) -- and it looks almost identical to the base usermanager with H2 / SQLite šŸ™‚

2024-09-27T03:28:49.092299Z

yeah, that reminds me--I think it was after I made that biff-XTDB2 example that the xt team announced they were gonna mostly just focus on the sql dialect going forward. so that repo is probably not even how I would do it today if I redid it, but it'll at least show you the parts that need to be changed. after XT2 is generally available I'll probably also see if using https://github.com/seancorfield/honeysql in the biff starter repo would be a good idea.

jf 2024-09-27T03:31:40.054289Z

ok. I would not want to go to SQL, quite frankly. I'm hoping that the new XTQL gets some love and support as well.

šŸŽ‰ 1
2024-09-27T03:38:40.385729Z

the new variant on sql the xt team is developing, possibly combined with honeysql, honestly seems very promising to me--I would at least check it out/wait and see how it turns out (I haven't been following along extremely closely so I can't say off the top of my head if they've already released most/all of the new syntax/grammar they've had planned or if it's still in the works...)

šŸ‘€ 2
āž• 1
jf 2024-09-27T03:49:21.646769Z

gotcha. Thanks for the note!

šŸ‘Œ 1
šŸ‘ŒšŸ¼ 1
Nik 2024-10-03T03:11:50.703349Z

I’m loving the current experience with biff. Im right now in experimental phase of a project I aiming to monetize in future. I’m interested once XT2 is ready and decide whether to go with PostgreSQL or XTDB for scale

šŸ™Œ 1