Socket interface answers/errors

[Lord-Evil]

E.g. user in a contact list with no history. We do sock.send("history username") but beeing blocked on reed because telegram-cli says nothing :-) Pleas add "ANSWER 0" as reply to whatever socket command if there's no data to reply.

[aneip]

I believe when we send msg (command 'msg') also return nothing. Can we have some status whether msg is send or not.

[Lord-Evil]

aneip, are you making remote client or local wrapper?

[vysheng]

Yep, I'll add some kind of answer.

[luckydonald]

Can we have replies when the CLI failed (eg. parsing the string)?
Right now we require a timeout to determine that.

[zyberspace]

Also, some commands return just a new line ("\n") and some the "ANSWER [Bytes]" string.
Why not "ANSWER 7" and then "SUCCESS" in those cases, like for example for "status_online"?

Parsing the answer would be much easier this way. 👍

[luckydonald]

There should be better (as in more) error messages.
Not only when using sockets, but generally.
For example:
dialjg_list should return/print something like "unknown command"
msg this_user_does_not_exist "test" something like "unknown peer"
msg user something like "missing parameter"
especially when used with sockets, it should always return something.
This issue gains importance with using json (#516) because this makes sockets useful again, and there output is expected.
@zyberspace
Yep. Socket answer parsing would be much easier without need for timeouts. (I didn't encounter linebreaks)

[luckydonald]

I did a fork/branch thingy gaining in that direction:

but sadly many things/arguments are parsed/validated only in the library, so getting them errors is difficult, and needs insight to the tgl, (and maybe refactoring?)

[luckydonald]

@Lord-Evil Can you please rename this thread to something more meaningful, like "Socket interface answers/errors"?

[zyberspace]

@Lord-Evil Can you please rename this thread to something more meaningful, like "Socket interface answers/errors"?

+1 👍

[Lord-Evil]

@luckydonald done
The most important error message to me seems to be needed for "load_*" commands. Cuz you're never know whether there is no such a file (e.g. wrong msg id) or file size is so big that needs to be awaited for dozens of minutes to download.

[vysheng]

Added some error codes