Hi All,

Since i'll be leaving Tamedia (who sponsored development of Asdf) and it's not under active use anymore here we decided to donate it to libmir.

Asdf has a few nifty features that make it perfect for reading large amounts of json lines and transforming data. For an example use case see je, the json to csv transformer app ( https://github.com/tamediadigital/je )

I still think that it is by far the easiest to use and fastest production grade Json Library for D.
Maybe it should be renamed to reflect that, proposals welcome.

cheers,
y

On Wednesday, 11 July 2018 at 21:42:45 UTC, yannick wrote:
> Hi All,
>
> Since i'll be leaving Tamedia (who sponsored development of Asdf) and it's not under active use anymore here we decided to donate it to libmir.
>
> Asdf has a few nifty features that make it perfect for reading large amounts of json lines and transforming data. For an example use case see je, the json to csv transformer app ( https://github.com/tamediadigital/je )
>
> I still think that it is by far the easiest to use and fastest production grade Json Library for D.
> Maybe it should be renamed to reflect that, proposals welcome.
>
> cheers,
> y

cool to hear that you don't let that library die. I like asdf but I never really had a usecase for it. When using vibe.d I just parse my json using their serializer. Additionally I never have the case of my input being line separated json, for example when reading a dub registry dump it's a 40MB json array where I don't really see application for asdf, or can it also be used with generic arrays that aren't line separated?

Anyway pretty cool library, would definitely use it if I ever see line separated json data in the wild.

On Wed, Jul 11, 2018 at 11:45 PM yannick via Digitalmars-d-announce < digitalmars-d-announce@puremagic.com> wrote:

> Hi All,
>
> Since i'll be leaving Tamedia (who sponsored development of Asdf) and it's not under active use anymore here we decided to donate it to libmir.
>
> Asdf has a few nifty features that make it perfect for reading large amounts of json lines and transforming data. For an example use case see je, the json to csv transformer app ( https://github.com/tamediadigital/je )
>
> I still think that it is by far the easiest to use and fastest
> production grade Json Library for D.
> Maybe it should be renamed to reflect that, proposals welcome.
>
> cheers,
> y
>
>
I have tried it few days ago and it does not work for me at  first. Because
I always read only "Simple Example".
One of my problem was that I have a struct with little different name that
was attribute name used in
json (positions and types has been correct), but it does not work and the
error was really wierd so it took
me a while until I have found out what was wrong.

Other problem was with types when one json attribute can be a number or null

On Wednesday, 11 July 2018 at 21:42:45 UTC, yannick wrote:
> Hi All,
>
> Since i'll be leaving Tamedia (who sponsored development of Asdf) and it's not under active use anymore here we decided to donate it to libmir.
>
> Asdf has a few nifty features that make it perfect for reading large amounts of json lines and transforming data. For an example use case see je, the json to csv transformer app ( https://github.com/tamediadigital/je )
>
> I still think that it is by far the easiest to use and fastest production grade Json Library for D.
> Maybe it should be renamed to reflect that, proposals welcome.
>
> cheers,
> y

Great! Thanks for keeping it alive.

I use the library specifically for reading json files stored on s3 that were generated by  aws' kinesis service. (The files have multiple json roots in them, but aren't separated by newlines; just one after another. I created a small custom parseJsonByLine for that).

On Wednesday, 11 July 2018 at 21:42:45 UTC, yannick wrote:
> Hi All,
>
> Since i'll be leaving Tamedia (who sponsored development of Asdf) and it's not under active use anymore here we decided to donate it to libmir.
>
> ...
>
> cheers,
> y

Thanks for keeping it alive. I've used asdf at work last week to build a json plugin for logparser:

https://en.wikipedia.org/wiki/Logparser

I have yet to ask permission to open source it, but I'll make an announcement if the boss agrees.

On Wednesday, 11 July 2018 at 22:21:32 UTC, Daniel Kozak wrote:

> I have tried it few days ago and it does not work for me at  first. Because
> I always read only "Simple Example".
> One of my problem was that I have a struct with little different name that
> was attribute name used in
> json (positions and types has been correct), but it does not work and the
> error was really wierd so it took
> me a while until I have found out what was wrong.
>
> Other problem was with types when one json attribute can be a number or null

Asdf was actually exactly created for this use case.
You can do transformations when reading and writing.
Also the null case is handled via different tools, please read the Readme and look into the API.

If there is an use case that you seem not to be able to pull of, please file a ticket.

On Wednesday, 11 July 2018 at 21:58:29 UTC, WebFreak001 wrote:
> cool to hear that you don't let that library die. I like asdf but I never really had a usecase for it. When using vibe.d I just parse my json using their serializer. Additionally I never have the case of my input being line separated json, for example when reading a dub registry dump it's a 40MB json array where I don't really see application for asdf, or can it also be used with generic arrays that aren't line separated?

i never understood why people do this. why would you ever generate such a big json, it requires a lot of memory to generate it too.
that being said asdf also works in this case and i think you should even be able to stream non th e less (not 100% sure, if not it could/should be implemented).

> Anyway pretty cool library, would definitely use it if I ever see line separated json data in the wild.

die Hoffnung stirbt zuletzt...

anyway, really happy to see that asdf found a few uses and i can really recommend to try it out, even to replace the standard json libs in e.g. vibe.d, which is what i do too.
Its not only faster but also easier, since you can use your structs/classes and have granular control over what properties are (de-)serialized.