Re: Add redis-db module to upstream?

[Sebastian Gniazdowski <psprint@zdharma.org>]

On 3 czerwca 2017 at 21:35:52, Daniel Shahaf (d.s@daniel.shahaf.name) wrote:
> This module was created 10 days ago.

It is based on my db/gdbm, which was based on original db/gdbm, so you have two ancestors in the path (there was also Zpopulator, but not released). 176 direct commits, not a weekend-coding with 20 commits.

> It has had no bug reports, no pull requests

No bug reports because like any my project it has more than 200 commits essentially (and I've used tests and Valgrind). E.g. ZUI is 555 direct commits, and probably additional 1000 from ancestors.

> and no code contributors other than its author.

Single-author is a good thing, if focused and engaged. I recall times of XFree86 where single hacker (the one who invented the hacker-sign from game of life?) did Damage extension and multiple others, essentially opening the way to Xorg, IMO. I now recall, it was Keith Packard, his extensions: XRender, XFixes, XDamage, XComposite, XRandR. Hah, it's thanks to him that one can call xrandr and just get current resolution.

Anti-example – universal-ctags has many contributors, but they don't support such a plain thing like multiple arguments to typeset/local/declare. It was a first thing that I've added to my ctags.

> Does it have
> any users besides Sebastian? Will anyone on this list use this module
> in his setup?

It has explicitly *strong* use case. On #redis I meet people who are using python, because redis-cli like most reference-clients is just an inadequate tool when there's actual work to do:

https://walrus.readthedocs.io/en/latest/containers.html#lists

So people open python-cli and manage database by programming. Zsh data structures are much more convenient, sorting, upper-casing etc., searching etc. available via flags, that's not-programming, it's more like just using.

-- 
Sebastian Gniazdowski
psprint /at/ zdharma.org