On 07/20/2016 12:15 PM, Jeff King wrote:
One final bit of food for thought. Just yesterday somebody asked me about renewing the old idea of using a more standardized format for machine-readable output, like --json. That's obviously something that would exist alongside the existing formats for compatibility, and it doesn't fundamentally change anything about adding a new format as your patches do (it just becomes yet another format). However I wanted to mention it in case you are intrigued by the idea, and would be interested in skipping porcelain-v2 entirely in favor of moving to something like json. A totally reasonable response is "haha no. Please stop moving the goalposts". I just wanted to throw it out there as an option (and in case you are interested, to let you think about it before any more work goes into this direction).
haha no.... :-) Short term, I'd rather nail down what I have now (both content-wise and format-wise) and see how we like it. And have a follow-up task to look at the --state header we spoke of earlier. And save the JSON version as an independent task for later. I understand the motivation for a JSON option (and have thought about it before) but I think it ought to be kept separate. At a higher-level, it seems like a JSON option would be an opportunity to start a project-wide conversation about formats, consistency, plumbing, and etc. A top-down conversation if you will about which commands will/won't get enhanced, legacy cruft that would not need to be converted, JSON style and naming and consistency issues, current best practices in the node/whatever community, and etc. I could be wrong, but this feels like a top-down feature conversation in a wider audience. Jeff -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html