Open AntonHermann opened 2 years ago
I don't think it actually waits for the end of the input file, it just reads the input in chunks.
So tail -f x.csv | xsv...
works, it just outputs stuff in 32kB chunks. You can see this when you use a file where data is rapidly appended to.
So it should be possible to build this without any large modifications by just reducing the buffer size, possibly here
While it's perfectly reasonable to require reading the whole input before processing it, as discussed in #192 for some other subcommands, like
xsv fmt
orxsv select
I see no obvious reason why they shouldn't be line buffered, although that seems to be the case.If there is some kind of technical limitation I'm overseeing, would a
--streaming
option or something alike be possible?