-
Notifications
You must be signed in to change notification settings - Fork 1.2k
cli - get - fix formatting when piping out #900
Conversation
@diasdavid heres a fix -- seems like circle tests failed due to unrelated chrome upgrade failure |
thanks @kumavis! Yeah circle has been doing that, we haven't had the time to investigate, but it is just how it is installing chrome. Going to wait for CI on this one. |
the test failure is interesting |
ill compare against go-ipfs |
my brain just exploded |
TIL hexdump respects machine endianness in its two-byte display this is better
|
this is surprisingly complicated because |
currently we also append a newline to the CID on a put
go-ipfs also does this
not sure what the unix conversions are here cat does not append a newline
curl does not apply a newline |
@whyrusleeping @Kubuxu wanna chime in here? |
per discussion in irc, we thought that:
|
somewhere i trolled myself and made some bad assumptions about go-ipfs behavior will take another look tomorrow and more closely compare go-ipfs + js-ipfs |
ok so the issue is that we:
this is complicated by the fact that we:
so we can:
|
@diasdavid @dignifiedquire pick your poison and I'll update the PR |
@kumavis why isn't the option to have a flag --no-newline (or something) not an option? Wasn't that the cleanest and obvious? |
@diasdavid i previously incorrectly stated go-ipfs behavior
|
yeah, let's match go-ipfs then :) |
btw, you can disable that misleading stripping of EOF from execa https://www.npmjs.com/package/execa#stripeof :) |
yeah, though it would mean we'd have to change all the tests to expect a newline |
|
the trifecta! 😱 |
closing in favor of #903 |
HTTP API does not need this and we should not ask people to set it if they don't really need it. License: MIT Signed-off-by: Marcin Rataj <[email protected]>
Fixes #897