Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
40 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Adrian Buehlmann
On 2012-08-09 21:57, Augie Fackler wrote:
> Yes, there is: the (hypothetical git) user will get angry that the
> branch is permanent, and decide Mercurial is a useless piece of trash
> and never look back.

I think you still haven't delivered any convincing argument why we
should care about such angry stubborn users B at the expense of annoying
current users A.

Also, I note again that B are users who obviously have refused to read
anything about Mercurial branches (specifically, they obviously applied
a command before having read 'hg help branch') and who seem to think
they can blindly apply concept X named Y from git on Mercurial.

What's more, your "shoot in the foot" argument is a complete red
herring. If you start using a tool that's supposed to store your data
but you refuse to read anything before using it, then you have clearly
done something wrong. Blame yourself.

I wouldn't be surprised if there are more concepts from git, which can't
be blindly applied to Mercurial without causing havoc. Do we now have to
study git and warn on a pile of other command usages as well? For
example, do we now have to warn that 'hg pull' is not going to do a
merge as git's pull does? No.

Mercurial is not git. It's as simple as that.

Trying to start making Mercurial look like git is a dead end (IMHO).
Even if it's only by assuming that someone who executed "hg branch" most
likely made an error (and telling them "hey, you most likely wanted
'bookmarks'!").
_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Adrian Buehlmann
In reply to this post by Adrian Buehlmann
On 2012-08-08 20:46, Adrian Buehlmann wrote:

> On 2012-08-08 18:44, Matt Mackall wrote:
>> On Wed, 2012-08-08 at 11:49 +0400, [hidden email] wrote:
>>> Hello!
>>>
>>> As I've already mentioned (several times :-D) in Bug 3321
>>> (http://bz.selenic.com/show_bug.cgi?id=3321),
>>
>> Indeed you have, and you got a number of definitive responses there. Why
>> should I even read this thread when it seems its point is simply to try
>> to gather popular support for a solution that I've already rejected? If
>> I were the type of project leader who accepted solutions I didn't like
>> just because they were popular.. we'd all be using Git.
>>
>> If you want to make forward progress, suggest a solution that hasn't
>> been definitively rejected. For instance, I don't know.. how about
>> automatically suppressing the message in repos that already have 10
>> branches? That's probably sufficiently n00b-proof, while getting out of
>> the way of heavy branch users.
>
> How about:
>
> diff --git a/mercurial/commands.py b/mercurial/commands.py
> --- a/mercurial/commands.py
> +++ b/mercurial/commands.py
> @@ -912,8 +912,8 @@
>                                       hint=_("use 'hg update' to switch to it"))
>              repo.dirstate.setbranch(label)
>              ui.status(_('marked working directory as branch %s\n') % label)
> -            ui.status(_('(branches are permanent and global, '
> -                        'did you want a bookmark?)\n'))
> +            ui.status(_('(the branch name will be permanently recorded '
> +                        'on commit)\n'))
>      finally:
>          wlock.release()
>
> Which looks like this:
>
>   $ hg branch foo
>   marked working directory as branch foo
>   (the branch name will be permanently recorded on commit)
>
> Instead of (currently):
>
>   $ hg branch foo
>   marked working directory as branch foo
>   (branches are permanent and global, did you want a bookmark?)

Apologies for filibustering (see my other replies in this thread), but I
just had an interesting in-person discussion about this off-list with
someone else. I think a somewhat interesting aspect that wasn't
mentioned yet came up during that discussion:

The current output of the command is (as we know, let me just repeat it
here again):

  $ hg branch foo
  marked working directory as branch foo
  (branches are permanent and global, did you want a bookmark?)

If that message ("did you want a bookmark?") would help those purported
git-refugees (who don't read help tests, apparently), then that hint
would be _too late_ anyway at that point.

The branch 'foo' has already been set when they see that "warning". If
they don't know what they do, they will commit anyway (and thus set the
branch). I don't think a user who doesn't read help texts before using
commands will be able to get rid of that branch.

Long story short: I think there is no alternative to reading help texts
(in this case).

_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Arne Babenhauserheide-2
In reply to this post by Adrian Buehlmann
Am Mittwoch, 8. August 2012, 20:46:53 schrieben Sie:

> How about:
>
> diff --git a/mercurial/commands.py b/mercurial/commands.py
> --- a/mercurial/commands.py
> +++ b/mercurial/commands.py
> @@ -912,8 +912,8 @@
>                                       hint=_("use 'hg update' to switch to
> it")) repo.dirstate.setbranch(label)
>              ui.status(_('marked working directory as branch %s\n') % label)
> -            ui.status(_('(branches are permanent and global, '
> -                        'did you want a bookmark?)\n'))
> +            ui.status(_('(the branch name will be permanently recorded '
> +                        'on commit)\n'))
>      finally:
>          wlock.release()
>
> Which looks like this:
>
>   $ hg branch foo
>   marked working directory as branch foo
>   (the branch name will be permanently recorded on commit)
I like it.

The nagging also disturbs me more every month, and your change looks good to
me.

It’s nice to cater to new users, but even more important is keeping the usage
of current users flow seamlessly.

Best wishes,
Arne
--
1w6 sie zu achten,
sie alle zu finden,
in Spiele zu leiten
und sacht zu verbinden.
http://1w6.org


_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel

signature.asc (325 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Arne Babenhauserheide-2
In reply to this post by Adrian Buehlmann
Am Mittwoch, 8. August 2012, 20:46:53 schrieben Sie:
>   $ hg branch foo
>   marked working directory as branch foo
>   (the branch name will be permanently recorded on commit)

PS: this actually gives two additional pieces of information:

- branch names are permanent
- they are recorded on commit

The one thing I miss is how to undo the marking.

Maybe like this:

    $ hg branch foo
    marked working directory as branch foo (previous name: bar)
    (the branch name will be permanently recorded on commit)

I’m not really happy with that, though. Too cluttered.

Maybe it helps someone find a better wording.

Information which should be in there:

- new branch name
- branches are permanent
- the branch name is recorded on commit
- old branch name (to be able to undo the change without looking up stuff)

Best wishes,
Arne
--
1w6 sie zu achten,
sie alle zu finden,
in Spiele zu leiten
und sacht zu verbinden.
http://1w6.org


_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel

signature.asc (325 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Arne Babenhauserheide-2
In reply to this post by harry
Am Donnerstag, 9. August 2012, 01:59:27 schrieben Sie:

> Laurens Holst wrote
>
> >> If that is unswallowable, a message that
> >> doesn't imply the user is doing something wrong would be better.
> >
> > So where are your concrete suggestions? Or do you expect someone else to
> > do that for you?
>
> How about: "The branch name will be permanently recorded on commit. For
> alternatives, see hg help bookmarks."
Maybe shorter:

"The branch name will be permanently recorded on commit. For
 alternatives, see hg help."

That should be future proof in case someone invents a new way of branching
someday. Also it includes feature clones and anonymous branching.

Best wishes,
Arne
--
Ich hab' nichts zu verbergen – hab ich gedacht:

- http://draketo.de/licht/lieder/ich-hab-nichts-zu-verbergen


_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel

signature.asc (325 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

harry
In reply to this post by Adrian Buehlmann
With regards to the question of why we should worry about git users, git seems to have become the de facto gold standard DVCS. If one model of car had some controls that differed from the expectations set by all popular models, you would expect some appropriate warning for new users, without relying on them to read the manual before doing anything. It is reasonable that if hg branch is substantially different from git branch, some warning should be given to new users.

Is it possible to refer new users to GitConcepts? A prominent link could be added to the download page, but what about those who download hg from their package repository? (I use Windows, so I'm entirely ignorant as to how Linux package repositories work.)
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Adrian Buehlmann
On 2012-08-12 09:16, v wrote:
> With regards to the question of why we should worry about git users, git
> seems to have become the de facto gold standard DVCS. If one model of car
> had some controls that differed from the expectations set by all popular
> models, you would expect some appropriate warning for new users, without
> relying on them to read the manual before doing anything. It is reasonable
> that if hg branch is substantially different from git branch, some warning
> should be given to new users.

Not if our car (Mercurial) had a different concept of what the term
"branch" means from its beginning.

I also don't really care what you claim what the "de facto" standard for
a DVCS is. There is a reason why I'm using Mercurial (and not git).
There is no reason to tell me I should use command X instead of Y if I
have chosen to use X. Unless X (=branch) is deprecated, which is clearly
not the case for Mercurial's branch command.

(If I would apply your - erroneous, FWIW - reasoning to Linux vs
Windows, then Linux would have to add a couple of messages to its UI
strings as well. No one really expects that, do you?).

> Is it possible to refer new users to GitConcepts?

In Mercurial's UI texts for when someone just uses the branch command? No.

But you could probably do that in the help texts.

> A prominent link could be
> added to the download page, but what about those who download hg from their
> package repository? (I use Windows, so I'm entirely ignorant as to how Linux
> package repositories work.)

_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Adrian Buehlmann
On 2012-08-12 10:38, Adrian Buehlmann wrote:
> There is no reason to tell me I should use command X instead of Y if I
> have chosen to use X. Unless X (=branch) is deprecated, which is clearly
> not the case for Mercurial's branch command.

I made a mess with X any Y. This should really read:

"There is no reason to tell me I should use command Y instead of X if I
have chosen to use X. Unless X (=branch) is deprecated, which is clearly
not the case for Mercurial's branch command."
_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Arne Babenhauserheide-2
In reply to this post by vitalif
Hi Adrian,

Am Sonntag, 12. August 2012, 08:44:46 schrieb Adrian Buehlmann:
> Thanks for your emails. It looks like you forgot to cc the list though.
> Please send them to the list too.

Argl, yes. Different shortcuts in different programs… I need to do something
about that…

Thanks!

Best wishes,
Arne

 

> On 2012-08-11 14:49, Arne Babenhauserheide wrote:
> > Am Mittwoch, 8. August 2012, 20:46:53 schrieben Sie:
> >> How about:
> >>
> >> diff --git a/mercurial/commands.py b/mercurial/commands.py
> >> --- a/mercurial/commands.py
> >> +++ b/mercurial/commands.py
> >> @@ -912,8 +912,8 @@
> >>
> >>                                       hint=_("use 'hg update' to switch
> >>                                       to
> >>
> >> it")) repo.dirstate.setbranch(label)
> >>
> >>              ui.status(_('marked working directory as branch %s\n') %
> >>              label)
> >>
> >> -            ui.status(_('(branches are permanent and global, '
> >> -                        'did you want a bookmark?)\n'))
> >> +            ui.status(_('(the branch name will be permanently recorded '
> >> +                        'on commit)\n'))
> >>
> >>      finally:
> >>          wlock.release()
> >>
> >> Which looks like this:
> >>   $ hg branch foo
> >>   marked working directory as branch foo
> >>   (the branch name will be permanently recorded on commit)
> >
> > I like it.
> >
> > The nagging also disturbs me more every month, and your change looks good
> > to me.
> >
> > It’s nice to cater to new users, but even more important is keeping the
> > usage of current users flow seamlessly.
> >
> > Best wishes,
> > Arne
--
singing a part of the history of free software:

- http://infinite-hands.draketo.de


_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel

signature.asc (325 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Arne Babenhauserheide-2
In reply to this post by Adrian Buehlmann
Am Donnerstag, 9. August 2012, 22:48:52 schrieb Adrian Buehlmann:
> What's more, your "shoot in the foot" argument is a complete red
> herring. If you start using a tool that's supposed to store your data
> but you refuse to read anything before using it, then you have clearly
> done something wrong. Blame yourself.

So that’s what I should tell a git user who wants to contribute to my project?

That’s exactly the attitude from git: “your fault, RTFM”.

Mercurial always was different, and that’s great! Let’s keep it the vcs which
just works - and helps you out, when you’re in a bind.

Best wishes,
Arne
--
Ich hab' nichts zu verbergen – hab ich gedacht:

- http://draketo.de/licht/lieder/ich-hab-nichts-zu-verbergen


_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel

signature.asc (325 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Adrian Buehlmann
On 2012-08-13 10:42, Arne Babenhauserheide wrote:
> Am Donnerstag, 9. August 2012, 22:48:52 schrieb Adrian Buehlmann:
>> What's more, your "shoot in the foot" argument is a complete red
>> herring. If you start using a tool that's supposed to store your data
>> but you refuse to read anything before using it, then you have clearly
>> done something wrong. Blame yourself.
>
> So that’s what I should tell a git user who wants to contribute to my project?

Yes. At some point they have to accept the fact that the term "branch"
has a different definition from what it has on git. It's not the job of
the UI texts of Mercurial to be tailored for git refugees who refuse to
read help texts before applying commands.

Assuming that a git user can blindly start using Mercurial without
learning the concepts, and by assuming everything will work the same as
in git, or Mercurial will warn, is a false attitude IMHO. That won't work.

[FWIW, you seem to contradict with yourself (judging from your other
emails).]

> That’s exactly the attitude from git: “your fault, RTFM”.

How is that relevant here?

> Mercurial always was different, and that’s great! Let’s keep it the vcs which
> just works - and helps you out, when you’re in a bind.

And what should we conclude from that?
_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Arne Babenhauserheide-2
Am Montag, 13. August 2012, 11:24:04 schrieb Adrian Buehlmann:

> On 2012-08-13 10:42, Arne Babenhauserheide wrote:
> > Am Donnerstag, 9. August 2012, 22:48:52 schrieb Adrian Buehlmann:
> >> What's more, your "shoot in the foot" argument is a complete red
> >> herring. If you start using a tool that's supposed to store your data
> >> but you refuse to read anything before using it, then you have clearly
> >> done something wrong. Blame yourself.
> >
> > So that’s what I should tell a git user who wants to contribute to my
> > project?
> Yes. At some point they have to accept the fact that the term "branch"
> has a different definition from what it has on git. It's not the job of
> the UI texts of Mercurial to be tailored for git refugees who refuse to
> read help texts before applying commands.
Users don’t read manuals. Just catering to the 0-20% who do is a low goal…

> Assuming that a git user can blindly start using Mercurial without
> learning the concepts, and by assuming everything will work the same as
> in git, or Mercurial will warn, is a false attitude IMHO. That won't work.

For starting to contribute to my project, it will work. That’s the point I
want to make: A user with git-experience is a user, too.

Just letting him run into the knife will not only hurt him and my project (he
is less likely to contribute the small fixes which often start bigger
contributions, if the tool hurts him), but it will also reflect badly on
Mercurial, because he will think that hg is hard to use - and tell his
friends.

Git is hard to use. Most accept that, when someone tells them.
Hg is different. It just works for most people.

> [FWIW, you seem to contradict with yourself (judging from your other
> emails).]

Not really. I want a different warning, but I like it that we have one.

Best wishes,
Arne
_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel

signature.asc (325 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Adrian Buehlmann
On 2012-08-13 11:35, Arne Babenhauserheide wrote:

> Am Montag, 13. August 2012, 11:24:04 schrieb Adrian Buehlmann:
>> On 2012-08-13 10:42, Arne Babenhauserheide wrote:
>>> Am Donnerstag, 9. August 2012, 22:48:52 schrieb Adrian Buehlmann:
>>>> What's more, your "shoot in the foot" argument is a complete red
>>>> herring. If you start using a tool that's supposed to store your data
>>>> but you refuse to read anything before using it, then you have clearly
>>>> done something wrong. Blame yourself.
>>>
>>> So that’s what I should tell a git user who wants to contribute to my
>>> project?
>> Yes. At some point they have to accept the fact that the term "branch"
>> has a different definition from what it has on git. It's not the job of
>> the UI texts of Mercurial to be tailored for git refugees who refuse to
>> read help texts before applying commands.
>
> Users don’t read manuals. Just catering to the 0-20% who do is a low goal…

You are theorizing up to a point where it becomes pretty meaningless.

>> Assuming that a git user can blindly start using Mercurial without
>> learning the concepts, and by assuming everything will work the same as
>> in git, or Mercurial will warn, is a false attitude IMHO. That won't work.
>
> For starting to contribute to my project, it will work. That’s the point I
> want to make: A user with git-experience is a user, too.

A git user is a user, sure. What does that tell us?

> Just letting him run into the knife will not only hurt him and my project (he
> is less likely to contribute the small fixes which often start bigger
> contributions, if the tool hurts him), but it will also reflect badly on
> Mercurial, because he will think that hg is hard to use - and tell his
> friends.

How on earth is applying 'hg branch' letting him "run into the knife"?

Normally, I would expect that a user who is using a tool which is new to
him or her would make some experiments and see where he or she gets. And
spend a minute or too reading help.

Expecting that Mercurial will just work exactly the same a git is just
arrogant.

> Git is hard to use. Most accept that, when someone tells them.
> Hg is different. It just works for most people.

Sure. But requiring that it works like git is wrong.

>> [FWIW, you seem to contradict with yourself (judging from your other
>> emails).]
>
> Not really. I want a different warning, but I like it that we have one.

But you acknowledge that warning users that they might have taken the
wrong command is just arrogant as well, right?

And why do we need a "warning"? There is nothing wrong with using the
branch command.

Just as a reminder: My proposal was to write:

  $ hg branch foo
  marked working directory as branch foo
  (the branch name will be permanently recorded on commit)

If I understood you correctly, you liked that. Now you seem to argue
differently.


_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Nikolaj Sjujskij
In reply to this post by Arne Babenhauserheide-2

> Am Donnerstag, 9. August 2012, 22:48:52 schrieb Adrian Buehlmann:
>> What's more, your "shoot in the foot" argument is a complete red
>> herring. If you start using a tool that's supposed to store your data
>> but you refuse to read anything before using it, then you have clearly
>> done something wrong. Blame yourself.
> So that’s what I should tell a git user who wants to contribute to my  
> project?
  Do lots of users need a separate branch when contributing? Mostly it's  
just a patch to be sent or few commits to be pushed to  
BitBucket/GoogleCode/wherever. Granted, git people boast about branching  
like mad, but in general branching seems to me more like a feature to be  
used by head developers, not regular contributors.

  How about something along those lines (if you want a reminder about  
previous branch, which is redundant in my opinion):

   $ hg branch foo
     working directory branch has been changed to foo from bar
     (the branch will be permanently recorded on commit, see `hg help  
branch`)
_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Arne Babenhauserheide-2
Am Montag, 13. August 2012, 14:05:04 schrieb Nikolaj Sjujskij:
>   How about something along those lines (if you want a reminder about  
> previous branch, which is redundant in my opinion):
>
>    $ hg branch foo
>      working directory branch has been changed to foo from bar
>      (the branch will be permanently recorded on commit, see `hg help  
> branch`)

That sounds good.

maybe switch the to and from, that sounds more like a natural progression:

   $ hg branch foo
     working directory branch has been changed from bar to foo
    (the branch will be permanently recorded on commit, see `hg help  
branch`)

Best wishes,
Arne
--
Konstruktive Kritik:

- http://draketo.de/licht/krude-ideen/konstruktive-kritik


_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel

signature.asc (325 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

harry
Proposal to moderate the branch warning

http://bz.selenic.com/show_bug.cgi?id=3590
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Martin Geisler-2
In reply to this post by Laurens Holst
Laurens Holst <[hidden email]> writes:

> Op 08-08-12 21:51, [hidden email] schreef:
>> I would FAR rather see this line entirely removed.
>
> Definitively rejected, so why bring it up *again*?

Because we might not agree with the rejection? I might be unusual, but I
don't subscribe to the idea that we cannot talk about and examine past
decissions when they feels like a mistake to some (obviously not to
everybody, hence the discussion).

>> If that is unswallowable, a message that doesn't imply the user is
>> doing something wrong would be better.
>
> So where are your concrete suggestions? Or do you expect someone else
> to do that for you?

There's no real "work" here -- patches to remove the message or to make
it conditional on a config setting are trivial to write. The work is to
discuss if the message is necessary and if it does more harm than good.

--
Martin Geisler

aragost Trifork
Commercial Mercurial support
http://aragost.com/mercurial/
_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Nikolaj Sjujskij
In reply to this post by harry
Den 2012-08-14 14:10:42 skrev v <[hidden email]>:

> Proposal to moderate the branch warning
>
> http://bz.selenic.com/show_bug.cgi?id=3590
  This seems redundant to me. Discussion takes place here, why open new bug  
for that?
_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Martin Geisler-2
In reply to this post by Arne Babenhauserheide-2
Arne Babenhauserheide <[hidden email]> writes:

> Am Mittwoch, 8. August 2012, 20:46:53 schrieben Sie:
>>   $ hg branch foo
>>   marked working directory as branch foo
>>   (the branch name will be permanently recorded on commit)
>
> PS: this actually gives two additional pieces of information:
>
> - branch names are permanent
> - they are recorded on commit
>
> The one thing I miss is how to undo the marking.

That is described in 'hg help branch', the -C option will reset the
branch name to what it was in the working copy parent changeset.

> Maybe like this:
>
>     $ hg branch foo
>     marked working directory as branch foo (previous name: bar)
>     (the branch name will be permanently recorded on commit)
>
> I’m not really happy with that, though. Too cluttered.

I think leaving out the old branch name is fine. If people want to
revert to it and don't know about 'hg branch -C', then they should be
able to look it up in 'hg log'.

Btw, I also think it's great that the new message mentions that the
branch name is recorded on commit -- that's an important detail.

> Maybe it helps someone find a better wording.
>
> Information which should be in there:
>
> - new branch name
> - branches are permanent
> - the branch name is recorded on commit
> - old branch name (to be able to undo the change without looking up stuff)
>
> Best wishes,
> Arne

--
Martin Geisler

aragost Trifork
Commercial Mercurial support
http://aragost.com/mercurial/
_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel
Reply | Threaded
Open this post in threaded view
|

Re: Why you don't want to allow disabling "branches are global and permanent, did you want a bookmark"?

Arne Babenhauserheide-2
In reply to this post by Adrian Buehlmann
Am Montag, 13. August 2012, 11:51:57 schrieb Adrian Buehlmann:
> Expecting that Mercurial will just work exactly the same a git is just
> arrogant.
>
> > Git is hard to use. Most accept that, when someone tells them.
> > Hg is different. It just works for most people.
>
> Sure. But requiring that it works like git is wrong.

That’s why I only want it to inform people when they do something they cannot
revert easily later on.

> >> [FWIW, you seem to contradict with yourself (judging from your other
> >> emails).]
> > Not really. I want a different warning, but I like it that we have one.
>
> But you acknowledge that warning users that they might have taken the
> wrong command is just arrogant as well, right?

It feels arrogant against *me*, so yes :)

> Just as a reminder: My proposal was to write:
>
>   $ hg branch foo
>   marked working directory as branch foo
>   (the branch name will be permanently recorded on commit)
>
> If I understood you correctly, you liked that. Now you seem to argue
> differently.

No, that’s exactly what I am arguing *for*.

I think we currently have 3 groups here:

(1) No info at all
(2) Info that branches are permanent
(3) Warn that bookmarks might be more appropriate

I want (2), but when I argument with people who want (1), that could be
interpreted as support for (3) - which it is not.

But my preferences are:

(2) > (3) > (1)

Rather a warning than no info, but rather an info than a warning.

> and don't know about 'hg branch -C'

I did not know that, nice! :)

Best wishes,
Arne
--
1w6 sie zu achten,
sie alle zu finden,
in Spiele zu leiten
und sacht zu verbinden.
http://1w6.org


_______________________________________________
Mercurial-devel mailing list
[hidden email]
http://selenic.com/mailman/listinfo/mercurial-devel

signature.asc (325 bytes) Download Attachment
12