'record' is too generic a name - perhaps switch to subcommands?

I think that's a good idea.

On Sun, 2008-02-24 at 23:23 +0000, James Westby wrote:
> Public bug reported:
>
> Hi,
>
> The name 'record' for a command is quite generic. Also it
> is the darcs term for 'commit'.
>
> I also think there is an existing plugin for bzr to add darcs
> style (per hunk) commit, which uses the 'record' name.
>
> Attached is a patch to change the command name to
> 'record-loom'. I didn't add any deprecation in, as with the
> plugin only a few days in the wild I doubt anyone is too
> used to it yet.

We don't generally qualify our other command names (we don't have
push-branch, or commit-tree); I'd rather we brainstorm (e.g. on the
bazaar list) about names than do this hacky-and-ugly name.

-Rob

--
GPG key available at: <http://www.robertcollins.net/keys.txt>.

Hi all,

I reported a bug on the loom plugin stating

  The name 'record' for a command is quite generic. Also it
  is the darcs term for 'commit'.

  I also think there is an existing plugin for bzr to add darcs
  style (per hunk) commit, which uses the 'record' name.

  Attached is a patch to change the command name to
  'record-loom'. I didn't add any deprecation in, as with the
  plugin only a few days in the wild I doubt anyone is too
  used to it yet.

and Robert replied

  We don't generally qualify our other command names (we don't have
  push-branch, or commit-tree); I'd rather we brainstorm (e.g. on the
  bazaar list) about names than do this hacky-and-ugly name.

so I am bringing the discussion here.

My arguments for the name would be

  1. It is clear that it is to do with looms for someone just seeing
     the command name.
  2. It doesn't use the term from another VCS to mean something quite
     different.
  3. The TODO of looms makes it sound like 'record' is a lower level
     command that will one day be superseded by a 'commit-loom' command
     or similar. (I may have got the wrong end of the stick here).
  4. It doesn't overlap with an existing command provided by another
     plugin.
  5. There are already 'revert-loom' and 'show-loom' commands, so this
     fits in well in my opinion.

Anyone else have any opinions? Robert, do you have another suggestion
for how we should handle this?

Thanks,

James

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

James Westby wrote:
> and Robert replied
>
> We don't generally qualify our other command names (we don't have
> push-branch, or commit-tree); I'd rather we brainstorm (e.g. on the
> bazaar list) about names than do this hacky-and-ugly name.
>
> so I am bringing the discussion here.

Is there a reason you don't want to bring this to the Bazaar list as
Robert suggested?

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHwxJj0F+nu1YWqI0RAvYFAJ9KHv6JMAR1sC+ev/C7yclHeUdExgCfVoLf
QgEpQpOR2XmcqkNzTnAUKrE=
=tY8k
-----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

James Westby wrote:
| Hi all,
|
| I reported a bug on the loom plugin stating
|
| The name 'record' for a command is quite generic. Also it
| is the darcs term for 'commit'.
|
| I also think there is an existing plugin for bzr to add darcs
| style (per hunk) commit, which uses the 'record' name.
|
| Attached is a patch to change the command name to
| 'record-loom'. I didn't add any deprecation in, as with the
| plugin only a few days in the wild I doubt anyone is too
| used to it yet.
|
| and Robert replied
|
| We don't generally qualify our other command names (we don't have
| push-branch, or commit-tree); I'd rather we brainstorm (e.g. on the
| bazaar list) about names than do this hacky-and-ugly name.
|
| so I am bringing the discussion here.
|
| My arguments for the name would be
|
| 1. It is clear that it is to do with looms for someone just seeing
| the command name.
| 2. It doesn't use the term from another VCS to mean something quite
| different.
| 3. The TODO of looms makes it sound like 'record' is a lower level
| command that will one day be superseded by a 'commit-loom' command
| or similar. (I may have got the wrong end of the stick here).
| 4. It doesn't overlap with an existing command provided by another
| plugin.
| 5. There are already 'revert-loom' and 'show-loom' commands, so this
| fits in well in my opinion.
|
| Anyone else have any opinions? Robert, do you have another suggestion
| for how we should handle this?
|
| Thanks,
|
| James
|

Well, bzr shelve solved it in a different way. It provides "bzr shelve" and "bzr
unshelve" and all other operations are sub-commands. "bzr shelf
list/delete/switch/show/upgrade".

It might be reasonable to use that for looms. I think Robert has already hooked
into "bzr switch", which takes care of up/down-thread. So "bzr loom create"
instead of create-thread.

I honestly haven't used it a lot, so I'm not sure what commands are used all the
time (shelve/unshelve) versus infrequently (shelf list, shelf delete).

Looking at the list of commands in "bzr help loom" it breaks down into:

~ loomify
~ record

~ create-thread
~ down-thread
~ up-thread
~ combine-thread

~ revert-loom
~ show-loom

record actually seems the odd-man-out in this case. And "commit-loom" seems to
be a closer match. Since all the other commands have either 'thread' or 'loom'
in their names.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHwxyVJdeBCYSNAAMRAqOdAJsFA8HeVKxaeFbqITIls7M86GXLRACgvBuc
qNpyneDsntSMIkvF+k5vZno=
=Fm3c
-----END PGP SIGNATURE-----

On Mon, 2008-02-25 at 13:52 -0600, John Arbash Meinel wrote:
> Well, bzr shelve solved it in a different way. It provides "bzr shelve" and "bzr
> unshelve" and all other operations are sub-commands. "bzr shelf
> list/delete/switch/show/upgrade".
>
> It might be reasonable to use that for looms. I think Robert has already hooked
> into "bzr switch", which takes care of up/down-thread. So "bzr loom create"
> instead of create-thread.
>
> I honestly haven't used it a lot, so I'm not sure what commands are used all the
> time (shelve/unshelve) versus infrequently (shelf list, shelf delete).

Record is used before every time you push (just like commit is used
before every time you push).

> record actually seems the odd-man-out in this case. And "commit-loom" seems to
> be a closer match. Since all the other commands have either 'thread' or 'loom'
> in their names.

Well, switch, merge, push, pull all also operate on looms and don't have
loom in their name.

record wasn't named after darcs record, nor the bzr plugin that adds a
record command. We have a bit of a first-in-first-served issue I guess.

Anyhow, I think the bzr record command should really decorate 'commit'
and be 'commit -i' or something, this would make it discoverable to
users that have it installed (via commit --help).

-Rob

--
GPG key available at: <http://www.robertcollins.net/keys.txt>.

On Mon, 2008-02-25 at 20:21 +0000, Paul Moore wrote:

> While I haven't used the extension at all yet, I'm certainly
> interested in it. I hate the hyphenated commands - I'd much rather
> have simple one-word commands. I quite like the idea for using
> subcommands like "bzr loom create".

I'm very open to doing this; I'd like the code for it to be tasteful
though, and in particular it should be as easy to write commands as it
is today:
class cmd_loom_create(..)

is the current way to create a command called loom-create.

So

class cmd_create(cmd_loom)

or something?

> As far as particular commands are concerned:
>
> * loomify - I really dislike this command, it doesn't convey to me at
> all what it is. I'd suggest "bzr loom init".

Well loomify takes over an existing branch and makes it a loom. 'bzr
loom init' I would expect other users to expect it to create a loom from
scratch; just like 'bzr init' (and in fact, bzr init --loom should be
supported, its a bug its not at the moment).

> * record - I've no real problem with this, but "bzr loom commit"
> seems to fit the scheme quite nicely.
> * xxx-thread - Why not "bzr thread XXX"?
> * revert-loom and show-loom - Could be "bzr loom revert" and "bzr loom show".

Primarily a lack of infrastructure in bzr lead to the current naming
convention I used.

> This reflects my preference for single word commands, and subcommands,
> over multi-word-commands. Others may feel differently.

Like I say above, I'm entirely open to hyphens vs spaces; as long as its
kept clear and understandable in the code and UI.

-Rob
--
GPG key available at: <http://www.robertcollins.net/keys.txt>.
>

On Mon, 2008-02-25 at 17:19 -0500, John Yates wrote:
> On Monday, February 25, 2008 Paul Moore wrote:
>
> > * loomify - I really dislike this command, it doesn't convey
> > to me at all what it is. I'd suggest "bzr loom init".
>
> In what way is "loomifying" a branch different from a possible
> "reconfigure --loom" ?

This is a good point; I don't know if reconfigure is well enough
pluggable to hook this in there today.

-Rob

--
GPG key available at: <http://www.robertcollins.net/keys.txt>.

On 26/02/2008, John Arbash Meinel <email address hidden> wrote:
> | My arguments for the name would be
> |
> | 1. It is clear that it is to do with looms for someone just seeing
> | the command name.
> | 2. It doesn't use the term from another VCS to mean something quite
> | different.
> | 3. The TODO of looms makes it sound like 'record' is a lower level
> | command that will one day be superseded by a 'commit-loom' command
> | or similar. (I may have got the wrong end of the stick here).
> | 4. It doesn't overlap with an existing command provided by another
> | plugin.
> | 5. There are already 'revert-loom' and 'show-loom' commands, so this
> | fits in well in my opinion.
> |
> | Anyone else have any opinions? Robert, do you have another suggestion
> | for how we should handle this?
> |
> | Thanks,
> |
> | James
> |
>
>
> Well, bzr shelve solved it in a different way. It provides "bzr shelve" and "bzr
> unshelve" and all other operations are sub-commands. "bzr shelf
> list/delete/switch/show/upgrade".
>
> It might be reasonable to use that for looms. I think Robert has already hooked
> into "bzr switch", which takes care of up/down-thread. So "bzr loom create"
> instead of create-thread.
>
> I honestly haven't used it a lot, so I'm not sure what commands are used all the
> time (shelve/unshelve) versus infrequently (shelf list, shelf delete).
>
> Looking at the list of commands in "bzr help loom" it breaks down into:
>
> ~ loomify
> ~ record
>
> ~ create-thread
> ~ down-thread
> ~ up-thread
> ~ combine-thread
>
> ~ revert-loom
> ~ show-loom
>
> record actually seems the odd-man-out in this case. And "commit-loom" seems to
> be a closer match. Since all the other commands have either 'thread' or 'loom'
> in their names.

I agree with John and James, commit-loom would be the obvious choice.
At least until someone adds a way to either hang it off commit, or do
space-separated subcommands.

--
Martin

There is 'record' plugin on Plugins page (which apparently is out of date but still listed there).
This plugin is alos provide record command as in darcs.

James Westby пишет:
> Hi all,
>
> I reported a bug on the loom plugin stating
>
> The name 'record' for a command is quite generic. Also it
> is the darcs term for 'commit'.
>
> I also think there is an existing plugin for bzr to add darcs
> style (per hunk) commit, which uses the 'record' name.
>
> Attached is a patch to change the command name to
> 'record-loom'. I didn't add any deprecation in, as with the
> plugin only a few days in the wild I doubt anyone is too
> used to it yet.
>
> and Robert replied
>
> We don't generally qualify our other command names (we don't have
> push-branch, or commit-tree); I'd rather we brainstorm (e.g. on the
> bazaar list) about names than do this hacky-and-ugly name.
>
> so I am bringing the discussion here.
>
> My arguments for the name would be
>
> 1. It is clear that it is to do with looms for someone just seeing
> the command name.
> 2. It doesn't use the term from another VCS to mean something quite
> different.
> 3. The TODO of looms makes it sound like 'record' is a lower level
> command that will one day be superseded by a 'commit-loom' command
> or similar. (I may have got the wrong end of the stick here).
> 4. It doesn't overlap with an existing command provided by another
> plugin.
> 5. There are already 'revert-loom' and 'show-loom' commands, so this
> fits in well in my opinion.
>
> Anyone else have any opinions? Robert, do you have another suggestion
> for how we should handle this?
>
> Thanks,
>
> James
>
>
>
>

I just renamed the record plugin into interactive. This name change comes from the fact that the plugin now adds a --interactive (and -i) option to commit, support for other commands should also land in that plugin.

The previously named record command (from the record plugin) is renamed into record-patch.

On Fri, 2008-03-07 at 12:25 +0000, Ali Sabil wrote:
> I just renamed the record plugin into interactive. This name change
> comes from the fact that the plugin now adds a --interactive (and -i)
> option to commit, support for other commands should also land in that
> plugin.
>
> The previously named record command (from the record plugin) is renamed
> into record-patch.
>

Great, thanks for doing that.

My objection to the "record" name in the loom plugin still stands
though.

Thanks,

James

On 07/03/2008, James Westby <email address hidden> wrote:
> On Fri, 2008-03-07 at 12:25 +0000, Ali Sabil wrote:
> > I just renamed the record plugin into interactive. This name change
> > comes from the fact that the plugin now adds a --interactive (and -i)
> > option to commit, support for other commands should also land in that
> > plugin.

That sounds great!

I'd love to have that in the core.

> My objection to the "record" name in the loom plugin still stands
> though.

Me too.

--
Martin