Conversation
|
Great!
I see two changes:
|
|
Yeah, there are the two changes you mention. I am also looking at the ways to adapt the ipfs mount tests. |
|
Just pushed the desired change for
Yes, this is correct. It's a bit annoying but the exact UX is hard to get right. We opted to try out cleanly separating daemon commands as requiring the daemon to be run explicitly. Not yet sure this is TRTTD, but it's an experiment. what are your thoughts?
Ah! This is not good. it should. Tried it and see this: Yet, the options look correct, and even show up in the help: @mappum any ideas? |
b6b7b66 to
0e633f3
Compare
|
I force pushed the branch and now the 2 changes are now in 2 different patches. The patch related to cmd block does not take into account your change to the output, but it will be easy to do. About cmds mount and daemon, maybe there should be a short option available in mount that starts the daemon like -s (for start) or -d (for daemon), but we can add that later. |
|
About "ipfs help", we are using egrep "^Usage: +ipfs" to see if its output looks good. I would expect at least something like "Usage: ipfs [options] [arguments]" at the top, otherwise it's not very clear if ipfs will launch its own shell (like when you launch "python") or if it expects commands as arguments after "ipfs". |
|
Also with ipfs2, the "ipfs help" output goes to stderr instead of stdout previously. |
I like that idea -- #346
Hm, this is a bit non trivial, given how we setup autogeneration. How's the following (from a4488ca):
Fixed in 3df1513. |
|
@jbenet yeah, it is much better now, thanks! |
@jbenet @chriscool (not to be merged into master) This is a hack to run sharness tests on th ipfs2 binary. Instead of compiling cmd/ipfs, it compiles cmd/ipfs2 and copies this into test/bin/ipfs. I thought this would be enough to pass the `basic-commands` test, but it's not. Although the output is fairly similar, the `ipfs version` test fails. ``` test (feat/test2) λ. diff version1 version2 1c1 < ipfs version 0.1.7 --- > ipfs version 0.1.5 ``` I'm not very experienced with `sh` scripting, so perhaps I'm missing a key ingredient or maybe misunderstanding the the tests are meant to work. Would like to get input on this. Thanks, @maybebtc
The ouput from "ipfs add" changed in ipfs2. With the changes in this commit, this output change doesn't prevent the test to pass. With TEST_NO_FUSE=1 test t0040 now passes on my Linux machine. License: MIT Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
The ouput from "ipfs block" changed in ipfs2. With the change in this commit, this output change doesn't prevent the test to pass. Test t0050 now passes on my Linux machine. License: MIT Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
The ouput from "ipfs help" changed in ipfs2. With the change in this commit, this output change doesn't prevent the test to pass. Test t0010 now passes on my Linux machine. License: MIT Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
0e633f3 to
7ab9051
Compare
@maybebtc @jbenet
The goal of this branch is to help make the sharness tests pass with ipfs2.
It can be merged into a test branch in jbenet/go-ipfs, so that we can see what happens on travis.
The first patch in this PR is 3368d29. It was made by @maybebtc to make the sharness tests use ipfs2.
The second patch make the output of some tests match the output from some ipfs2 commands. It's possible we should instead change the output from the ipfs2 commands to match the tests, or that we should change both the tests and the output from ipfs2.