Feedback on vis5 spec

[ahincks]

It looks good! I've suggested some minor edits in the file. Here are some additional comments/questions.

• Should fpga_count be optional? And does the name fpga_count need to be ‘strongly recommended’, or should that advice just be CHIME/HIRAX specific?
• For the input axis, would it make sense for chan_id and correlator_input to have more flexible typing? For instance, you might imagine an experiment where the channel ID is alphanumeric (e.g., A01), or conversely, where the correlator_input is strictly numeric. But maybe allowing this flexibility would be inefficient in the reader code?
• Regarding the question marks under the stack axis:
  • I don't have a strong opinion about whether the reverse axis should be required or recommended.
  • I'm not sure what the ‘entries that aren't stacked’ refers to. Maybe stacks that have varying weights? My vague memory of this is that adding another entry called weight doesn't solve this, though.
• Should flags/input be changed to flags/input_weight?

[jrs65]

Thanks Adam! I'll go over these, although I'll take your opinion about whether data should still be plural under advisement.

[jrs65]

It seems the OED (screenshot) thinks that in computing contexts it is standard to use data as a singular mass noun. However, I think the A&A and MNRAS style guides both say that data should be plural, whereas ApJ seems to express no opinion and defers to Merriam-Webster who say that both forms are fine but imply that the only way data as plural survives in print is because of enforcement by style guides (https://www.merriam-webster.com/dictionary/data#usage-1).