subcommands for services, developer sites, AIs, Misc, search engines and --mail

[PaavoTakala]

I think the above things should be subcommands instead of flags. eg. ggl i "a flycatcher"

Pros:

• less characters to type
• loss of confusion (often -ab means -a -b)

Cons:

• loss of backwards combatability

Googling subcommands would still be possible, if there's a space after it. eg. ggl "i " (Currently you have to do the same when Googling flags.)

Should I start making this or are you opposed to the idea?

[taraqfarhan]

Well, how about considering a middle-ground approach to this: we keep the current flags but add short aliases as subcommands optionally.

That way ggl i "query" and ggl -i "query" both work. We'll get the ergonomic benefit without losing backward compatibility.

The -ab confusion remains for users still using the flag syntax, but anyone adopting the new subcommand style gets a cleaner experience by default.

Also, on the Googling flags concern, I think currently we don't need to add that extra space like ggl "i " to google the flags. We can do that normally just like ggl "i". (I just tested this myself on my device)

What do you think?

[PaavoTakala]

I think that sounds good.

Subcommands aren't implemented yet, so ggl "i" and ggl "i " work the same for now. However, ggl "-i" and ggl "-i " produce different results. (The former opens Google search and the later gives the following error: unrecognized arguments: -.)
A better example would probably be ggl "-am" and ggl "-am ". The former opens Apple Maps and the later Googles "-am".
Someone who forgot whether am meant before or after midday might run ggl "am". If subcommands would be implemented, that would just open Apple Maps instead of Googling "am".

So even if I don't remove subcommands, I would still technically be making a small backwards incompatible change. In practice it wouldn't be a big problem. The README could just instruct everyone to put a space before the closing quote, if the query doesn't already contain a space.

[taraqfarhan]

How about an alternative like a -- separator (ggl -- "am") which is a Unix convention for "everything after this is a literal argument." I think, the -- approach is cleaner than a trailing space. What's your opinion?

[PaavoTakala]

I, too, think that -- is cleaner, but it's also 2 characters longer. I think that both options should be available so you can choose which one to use. Kind of like short and long versions of flags.

Also, should the program throw an error if there is an argument before -- that doesn't have spaces and that's not a valid subcommand or flag? Or should the program just assume its part of the query? The latter might make the addition of a flag or a subcommand always backwards incompatible.

[taraqfarhan]

I too agree that both options should be available. And I think the program should throw an error if there is an argument before -- that isn't a valid flag or subcommand - that way, adding new flags/subcommands later won't retroactively change how old queries are interpreted.