I have created a posqtgresql client for zig with support for zig stable in dev-zig-stable branche
and zig upstream in dev-zig-upstream
how can I make it std-like in its style and code structure?
1 Like
I haven’t reviewed your project, but I’d say that’s a question best answered by yourself and applied during design and implementation rather than after the fact.
In general:
- Format the code using the Zig formatter
- Make your containers unmanaged: don’t embed allocators.
- Use the documented naming conventions
- Prefer DOD over OOP
However “being like std” really shouldn’t be the primary hoal. The primary goals should be simplicity, clarity, and efficiency in the applicable domain.
2 Likes
PostgresQL is crazy work on the capitalization!
you can see an example dir if you want to see it in action, no need for an actual review 
The API accepts an allocator in its functions with an internal allocator only used for the reader’s buffer
Yeah thanks, I have value.zig and Message.zig separated from the rest. but having less allocations is the goal
“PosthreSQL“ right?
Recently I have revisited my EDSM stuff
and, as a part of the activity, I decomposed connector machine into two “layers“:
- L1 - pure TCP-connector
- L2 - service specific (PG, AMQP, MQTT etc) layer
Source (C, but I do not see no problem to translate that into Zig) attached.
Note that
- it’s able to use
MD5 or SCRAM-SHA-256auth methods only, add another if you want, it’s pretty simple - there is no I/O in the code, only MEM/CPU bound stuff, I/O is a task for TX/RX machines
pgcx.c (13.7 KB)