Re: [PATCH] vnlog support

[Dima Kogan]

Tim Rice <trice@posteo.net> writes:

> Dima, I admit I am a bit hesitant about the thought that datamash
> should explicitly support an additional data format. Especially when,
> as far as I know, vnlog format is neither widely used in industry or
> academia, nor is it associated with a GNU project.
>
> Also, I don't see any replies Assaf made to the list which would imply
> that he intended to proceed with this. I think that if he was planning
> to accept a vnlog patch, he would have posted a reply to the mailing
> list indicating his support. Not that I feel obliged to agree with
> everything Assaf said; but if he had posted a reply to the mailing
> list, I would at least consider what he had to say.
>
> If explicit format support were of interest to datamash users, my
> priorities would be to check on the status of datamash's csv handling
> (and maybe add something like libcsv integration if the current
> implementation is lacking), and/or add json (eg cJSON integration.)
> Then again, we don't need to reinvent jq, so I would also hesitate
> about json.

Hi.

I guess I don't see a downside to supporting more formats. datamash
should support vnlog and csv because 99% of the work is already done and
because this would be beneficial to more people, and would attract more
users. datamash already has a large test suite, so there isn't a lot of
risk of breaking stuff.

vnlog is a part of all my public projects (mrcal in particular), and
internal stuff at JPL (not all me). I'm going to maintain my own patched
datamash for my own use if we decline to merge this, but I'd rather not
have to do that.

I guess the emails with Assaf weren't Cc-ed to the list. I can send them
over if you like, but they don't really matter: the current maintainer
can make his own decisions.

And independent of datamash: if you're the sort of person that does data
processing on the commandline, try out vnl-filter and feedgnuplot. I
promise you'll like them :)