std.data.json formal review (page 26)

October 09, 2018
Re: std.data.json formal review
Posted by Jonathan M Davis
in reply to Nicholas Wilson
Permalink
Jonathan M Davis
Posted in reply to Nicholas Wilson
Permalink
On Tuesday, October 9, 2018 5:45:02 PM MDT Nicholas Wilson via Digitalmars-d wrote:
> On Tuesday, 9 October 2018 at 18:07:44 UTC, Márcio Martins wrote:
> > On Tuesday, 28 July 2015 at 14:07:19 UTC, Atila Neves wrote:
> >> Start of the two week process, folks.
> >>
> >> Code: https://github.com/s-ludwig/std_data_json
> >> Docs: http://s-ludwig.github.io/std_data_json/
> >>
> >> Atila
> >
> > Sorry for the late ping, but it's been 3 years - what has happened to this? Has it been forgotten?
> >
> > Working with JSON in D is still quite painful.
>
> I presume it became vibe.data.json, there is also asdf if you're looking for some other library.

As I understand it, it was originally part of vibe.d (though I think that it may have been internal-only to be begin with) and was put into std_data_json for the attempt to get into Phobos. The version inside of vibe.d has continued to be maintained while the separate version hasn't really been. Either way, there wasn't enough agreement about the design during the Phobos review process for it to make it into Phobos, and Sonke gave up on it.

I've used std_data_json on a few projects, and it works reasonably well for reading JSON, but I've found it rather frustrating when writing JSON, because you have no control over the order it writes data in - which results in perfectly valid JSON, since the key-value pairs are not ordered, but it's really annoying to use it for configuration files and the like where you organize the file the way you'd like, and then your program completely reorders it when it needs to make an adjustment to the file. However, I really need to check out the properly maintained version in vibe.d, and as you say, there are other JSON parsers on code.dlang.org such as asdf. Writing a JSON parser is pretty easy. It's coming up with an API that would get through the Phobos review process that's not necessarily easy.

While we would like to replace std.json, someone is going to have to put in the effort to write (or complete) something - and push it through the Phobos review process - in order to replace std.json. And while there's clearly some interest in having certain modules in Phobos replaced (or in some cases, new modules added), there doesn't seem to be many people willing to push their code through the Phobos review process at this point even if they've put the time and effort into writing the code. They're far more willing to just put it up on code.dlang.org and leave it at that. I think that the fact that code.dlang.org works as well as it does has to a large extent killed off interest in attempting to put anything through the Phobos review process. It's been quite some time since anyone has made the attempt.

Personally, I don't think that we even need some of the stuff in Phobos that's in there (like JSON or XML parsers) and that having it on code.dlang.org makes more sense, but I do think that having subpar versions of them in Phobos is a problem - arguably even more so when we say that at the top of the documentation and have had for years as is the case with std.xml. I don't think that std.json is rated as badly, but it's been talked about as needing replacement for years, and std_data_json would have replaced it had it made it through the review process. So, we should probably replace it with something one of these days, but of course, someone has to put in the time and effort, which no one wants to do.

- Jonathan M Davis
Forums
