<isd>
So, tbh I would have been astonished if the meteor devs were open to maintaining support for mongo that old. I think we're probably on our own here.
<ocdtrekkie>
I agree, though I think it's worth highlighting that only supporting SSPL databases is a pretty big issue for open source projects like ours.
<ocdtrekkie>
Ultimately we may find a workaround that but it'll still be a problem for them.
<TimMc>
I'd be curious to hear the context on this.
<ocdtrekkie>
I will have to read those, heh. I am supportive of SSPL personally but I think it's challenging while the OSI is considered as authoritative to ship SSPL stuff. Too many people literally call it "proprietary" just because the OSI said no.
<isd>
Meh, I understand the the push to say "not OSI" doesn't count; SSPL is is more defensible, but usually when some company calls their product "open source" but is on some other license, it's something that's blatantly not -- e.g. cockroachdb has something that is more directly "you can't make a competing product"
<isd>
better than a full-on ms style EULA, but I get the worry about watering down the notion.
<kentonv>
can we just write a Mongo emulator based on sqlite and migrate to that? How hard could it be? (half-joking)
<isd>
I mean, that's almost FerretDB, but it's postgres instead of sqlite.
<isd>
Doesn't have oplog yet though, so not ready for us
<kentonv>
I think the oplog only matters if you have multiple clients
<isd>
Doesn't meteor's subscription stuff depend on that?
<kentonv>
(where the "client" is the Meteor app)
<kentonv>
I think if you have only one instance of the Meteor app, Meteor doesn't actually need the oplog? Or at least they didn't originally.
<kentonv>
I mean, originally Meteor didn't use the oplog at all, just did in-memory stuff, and it was a problem if you had multiple instances. I could totally imagine they've grown dependent on it since then, though.