threads / discuss / 1868

[Terminology] Push vs. upload?

Subject: [Terminology] Push vs. upload?

## tl;dr

6 messages between Sep 20, 2005 and Sep 21, 2005.

replies: 5people: 3as markdown or json

Petr Baudis· Sep 20, 2005, 00:11 UTC · lore

  Hello,

  I've realized today that I "forgot" to rename cg-push to cg-upload, so
I impulsively did so right away, but now I'm seeing that the core git
still has git-push and no git-upload. This is confusing and appears to
be quite inconsistent - push should be the opposite of pull, but pull is
defined as fetch+merge, while push does just the anti-fetch ("upload")
part. Yes, it does the subset protection, but git-fetch with remotes/
seems to by default do the superset protection as well.

  How should we call it, then?

-- 
				Petr "Pasky" Baudis
Stuff: http://pasky.or.cz/
VI has two modes: the one in which it beeps and the one in which
it doesn't.

Linus Torvalds· Sep 20, 2005, 00:55 UTC · re: Petr Baudis · lore

Re: [Terminology] Push vs. upload?

On Tue, 20 Sep 2005, Petr Baudis wrote:

Show 5 quoted lines

> 
>   I've realized today that I "forgot" to rename cg-push to cg-upload, so
> I impulsively did so right away, but now I'm seeing that the core git
> still has git-push and no git-upload. This is confusing and appears to
> be quite inconsistent - push should be the opposite of pull

No. There is no opposite of pull. You can't update the tree on the other end. It's fundamentally an unsafe operation - the other end is by definition not a private tree.

(Or, if you want to see it another way: "pull" is "pull"s own opposite: "ssh other-end + git pull from that end" ends up being what you mean.)

		Linus

Petr Baudis· Sep 20, 2005, 01:44 UTC · re: Linus Torvalds · lore

Re: [Terminology] Push vs. upload?

Dear diary, on Tue, Sep 20, 2005 at 02:55:08AM CEST, I got a letter where Linus Torvalds <torvalds@osdl.org> told me that...

Show 15 quoted lines

> 
> 
> On Tue, 20 Sep 2005, Petr Baudis wrote:
> > 
> >   I've realized today that I "forgot" to rename cg-push to cg-upload, so
> > I impulsively did so right away, but now I'm seeing that the core git
> > still has git-push and no git-upload. This is confusing and appears to
> > be quite inconsistent - push should be the opposite of pull
> 
> No. There is no opposite of pull. You can't update the tree on the other 
> end. It's fundamentally an unsafe operation - the other end is by 
> definition not a private tree.
> 
> (Or, if you want to see it another way: "pull" is "pull"s own opposite:  
> "ssh other-end + git pull from that end" ends up being what you mean.)

Yes, but I'm just arguing about the naming - if you just take the word "pull", "push" is the obvious opposite to that and vice versa, so if we are going to use words "pull" and "push" in the same context, this relation should be preserved. So "push" really _should_ mean

	ssh other-end 'git-pull'

whether it's a safe operation or not - if not, then GIT should have nothing called "push". But the current "push" is really an opposite of "fetch", so it should be called "upload", I believe.

-- 
				Petr "Pasky" Baudis
Stuff: http://pasky.or.cz/
VI has two modes: the one in which it beeps and the one in which
it doesn't.

Linus Torvalds· Sep 20, 2005, 01:59 UTC · re: Petr Baudis · lore

Re: [Terminology] Push vs. upload?

On Tue, 20 Sep 2005, Petr Baudis wrote:

Show 7 quoted lines

> 
> Yes, but I'm just arguing about the naming - if you just take the word
> "pull", "push" is the obvious opposite to that and vice versa, so if we
> are going to use words "pull" and "push" in the same context, this
> relation should be preserved. So "push" really _should_ mean
> 
> 	ssh other-end 'git-pull'

Well, I actually called it "send" as in "git-send-pack".

But anybody who has ever used BK would still type "push", so I'd make that an alias anyway.

Not "upload". Upload is the reverse of "download", and we don't do that.

		Linus

Petr Baudis· Sep 21, 2005, 16:52 UTC · re: Linus Torvalds · lore

Re: [Terminology] Push vs. upload?

Dear diary, on Tue, Sep 20, 2005 at 03:59:15AM CEST, I got a letter where Linus Torvalds <torvalds@osdl.org> told me that...

Show 17 quoted lines

> 
> 
> On Tue, 20 Sep 2005, Petr Baudis wrote:
> > 
> > Yes, but I'm just arguing about the naming - if you just take the word
> > "pull", "push" is the obvious opposite to that and vice versa, so if we
> > are going to use words "pull" and "push" in the same context, this
> > relation should be preserved. So "push" really _should_ mean
> > 
> > 	ssh other-end 'git-pull'
> 
> Well, I actually called it "send" as in "git-send-pack".
> 
> But anybody who has ever used BK would still type "push", so I'd make that 
> an alias anyway.
> 
> Not "upload". Upload is the reverse of "download", and we don't do that.

All right, now I agree that "upload" isn't that appropriate, since git-ssh-upload isn't actually for any push-related activity, even though it is a git-ssh-fetch counterpart - it is so in a different sense.

I don't want to name it cg-send because I think that would be way too easily confused with sending patches over e-mail. So it ended up being cg-push again. :-)

-- 
				Petr "Pasky" Baudis
Stuff: http://pasky.or.cz/
VI has two modes: the one in which it beeps and the one in which
it doesn't.

Junio C Hamano· Sep 20, 2005, 01:54 UTC · re: Petr Baudis · lore

Re: [Terminology] Push vs. upload?

Petr Baudis <pasky@suse.cz> writes:

> ... push should be the opposite of pull, but pull is defined
> as fetch+merge, ...

It is my recollection that this exact "shouldn't push be opposite of pull" issue was raised when the list had the glossary discussion. My impression was that the discussion settled on what we have right now...