It (fix-Win10 branch) also fails on my other Win 10 machine $ curl localhost:15000 % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 0 0 0 0 0 0 0 0 --:--:-- 0:00:01 --:--:-- 0curl: (7) Failed to connect to localhost port 15000: Connection refused
Is only mock supported for now for ‘safe-authenticator-cli’? On my Linux Mint 19.1 with Rust 1.36.0 I get the following error when doing ‘cargo build’:
error[E0432]: unresolved import `routing::Client`
12 | use routing::Client as Routing;
| ^^^^^^^^^^^^^^^^^^^^^^^^^^ no `Client` in the root
The ‘cargo build --features mock-network’ works (did test that already a while back).
A detail: with an interactive login, you have to first type in the secret, but in the example my.config.json on the README.md I see first the password. Or is it to illustrate that the order of arguments in the json config file doesn’t matter
after ’cargo run --features mock-network -- --daemon 41805safe auth’ works, but
$ safe keys create --test-coins --preload 32.45
[2019-08-10T15:41:05Z ERROR safe] safe_cli error: ‘–test-coins’ flag is not supported since it’s only available for “mock-network” feature
-> What am I doing wrong here?
other commands like ‘safe files’ do work for me
In the output of the ‘safe help cat’-cmd the following typo: …serlialisation…
I was expecting that safe cat worked like ls when given a ‘subdir’ as <location>, but that is of course not like that. For example ‘safe cat safe://mywebsite/contact/’ gives an ‘No data found…’-error, although safe://mywebsite/contact/form.html exists. Although that can be solved by introducing ‘safe ls’ (that does the necessary filtering, depending on the subdir).
Correct, non-mock-network will be supported soon whenever we finish the integration with new upcoming version of SCL
Test-coins works only when you run safe_cli with --features mock-network so before you run your command $ safe keys create --test-coins --preload 32.45 make sure you are using a binary built with mock-network feature.
Thanks! I’ll fix the typo in next PR. I think what you suggest are very valid suggestions, although I just tried cat on a local folder in my linux and it doesn’t work like that but just says: cat: tests/: Is a directory
The GNU cat commando in Linux indeed doesn’t work for directories.
The confusion is that the main safe site also seems to be a directory for people not familiar with the Safe RDF concepts (Edit: I mean FilesContainer probably). So in that regard it is not consequent -> or
‘safe cat safe://mywebsite’ and
‘safe cat safe://mywebsite/subdir’
both work or both don’t work.
True, totally agree, so yes perhaps as you say we need to have cat to list folders, or not do it for root location. Thanks a lot @draw I’ll bring this for discussions with other folks and also try to see where it fits in with priorities/plans to adjust it.
Not a high priority I think: can be added later, maybe by others. But ‘ls’-functionality (also for subdirectories), be it in the ‘safe cat’ or in a seperate 'safe ls'-commando makes for example auto complete functionality easier when scrolling through a ‘safe://…’-site without the need to downloading the site completely.
And what to do if you want to download a complete Safe site. For now ‘safe cat > .../correct/file’ for each file? And implementation of ‘safe get --recursive ...’ or ‘safe sync --recursive ...’ in the other direction maybe for later?
Ps: the 'safe keys create --test-coins --preload 32.45’ works now. Thank you for the instructions!
Just for anyone reading this thread, with @JPL we found the root cause of the issue he was having on Windows 10, and the problem is actually the lib we currently use in safe_auth CLI to prompt the user to enter secret/password doesn’t work well in MinGW (which is what JPL was using). Thus, if using other shell like windows cmd, it works just fine.
Can the CLI be tweaked so it’s authenticated using the safe browser (mock) bundle rather than safe-authenticator-cli or alternatively can the browser read the local files created using safe-cli? Would be nice to be able to view my ‘hello world’ website.
To be complete: where it will remain different from the ‘ls’ command (not that it is a problem), if I am correct, is that resembles more ‘ls --recursive …’: it also shows the contents of the ‘subdirectories’.