On Tuesday, 29 June 2021 at 18:25:52 UTC, someone wrote:

Not who you were asking but: they just don't work.

Maybe they don't support Windows for one reason or another.

Maybe their abstraction for whatever reason isn't able to handle basic types for the database backend (true story).

Maybe they have weird, non-nonsensical errors, or for some reason just can't connect when nothing else is wrong.

And maybe, sometimes they decide to just crash for random reasons. e.g. I tried to use the Hunt database library recently, and it's dtor would crash because it was missing a null check or something.

So I (and likely others) decide to just write our own. In my own case I just wanted to run super simple queries with super simple results, so made a very tiny wrapper around postgres' C library: https://github.com/BradleyChatha/dubstats2/blob/master/d/database/source/database.d

I used a tool called DStep to automatically convert the headers I needed over to D, and with a few manual tweaks, everything just worked. No hassle, no random shenanigans, I could actually start the project instead of wrestling with 20 different libraries.

There's a comment in the top of that file that describes my general feelings towards most dub libraries.

If you don't stick with the top libraries, you're basically in a barren wasteland in terms of "will this shit even work?" and "I could've just written my own version in the hours/days it's taken me to get this damn thing to do anything".

On Tuesday, 29 June 2021 at 23:32:21 UTC, SealabJaster wrote:

Doesn't matter, you're welcome :)

I would love to see D as a first-class citizen on postgreSQL like C ... just dreaming of. My first impressions so far (which I will eventually post here) are in the line of: terrific language, really-powerful, a joy to code with. So dreaming of seeing it integrated to something like postgreSQL is like ... well, faaantastic. I can imagine writing very sophisticated stored-procedures right within-the-database. Man, a killer feature, well-ahead of C.

What do you think of the possibility of making a coordinated effort to tackle a definitive pglib binding to D providing we start with very-well defined requirements and not going farther than needed to begin with ? I think we could start with x basic access, then make something intermediate access (not even touching the basic one), then advanced, whatever. I mean, not leaving the users of basic access spoiled when we are working a few steps higher and making the development mistakes every component will bring in. Just thinking.

Just seen it. When I stop coding for a while (I mean, learning D with my test-bed demo -not even a project yet), I use to browse the forum more-or-less in random order to take a glimpse of how things work (or not) in D-land; and I encountered lots of complains like you say for dub libraries and the like, so I suppose taking away some very few major-ones they tend to be poor quality at the least, not a complain of mine, to be clear, just stating what I am seeing.

Top-libraries or else ... crystal-clear.