Differences between revisions 23 and 24
Revision 23 as of 2012-01-23 16:56:37
Size: 7606
Comment:
Revision 24 as of 2012-01-23 17:22:24
Size: 8635
Comment:
Deletions are marked like this. Additions are marked like this.
Line 45: Line 45:
A changeset is not expected to automatically move from a lower phase to an higher phase (eg: from ''public'' to ''draft'') but automatic '' '' A changeset is not expected to automatically move from a lower phase to an higher phase (eg: from ''public'' to ''draft'') but automatic
Line 48: Line 48:
Phase movement are automatic and transparent and most user don't have to care much about them. The base rule is very simple: '' '' Phase movement are automatic and transparent and most user don't have to care much about them. The base rule is very simple:
Line 50: Line 50:
 . “''Any changesets on a remote repository is seen a public''” '' ''  . “''Any changesets on a remote repository is seen a public''”
Line 52: Line 52:
On standard exchange commands, the phase of changesets on both side are compared. If differ the lowest phase is choosed. (eg: a changeset known as ''draft'' locally but '''public''' remotely is set public localy. Because public < draft) '' '' On standard exchange commands, the phase of changesets on both side are compared. If differ the lowest phase is choosed. (eg: a changeset known as ''draft'' locally but '''public''' remotely is set public localy. Because public < draft)
Line 54: Line 54:
This update happen during standard exchange commands: '' '' This update happen during standard exchange commands:
Line 56: Line 56:
* '''pull''': remote phase data are used to update the phase data on '' '' * '''pull''': remote phase data are used to update the phase data on
Line 58: Line 58:
 . the local repo. As pull is read only, it does not change changeset's phase on the remote '' ''  . the local repo. As pull is read only, it does not change changeset's phase on the remote
Line 60: Line 60:
* '''push''': remote phase data are used to the phase data on the local repo and '' '' * '''push''': remote phase data are used to the phase data on the local repo and
Line 62: Line 62:
 . then local phase data are pushed to the local repo. '' ''  . then local phase data are pushed to the local repo.
Line 64: Line 64:
The real behavior is a it's a bit more complicated than '' changesets on a remote repository is seen a public'', but this is true for repository keeping default . If you need a finer behavior, consult the ''[[#Publishing_Repository|publishing repository]]'' section. '' '' The real behavior is a it's a bit more complicated than '' changesets on a remote repository is seen a public'', but this is true for repository keeping default . If you need a finer behavior, consult the ''[[#Publishing_Repository|publishing repository]]'' section.
Line 66: Line 66:
New changeset committed locally are ''draft'', but some extension like ''mq'' may create ''secret'' and handle the move from ''secret'' to ''draft'' automatically '' '' New changeset committed locally are ''draft'', but some extension like ''mq'' may create ''secret'' and handle the move from ''secret'' to ''draft'' automatically
Line 68: Line 68:
Consult the [[#upgrade_Notes]] section to check how phase will move the first time a new version of Mercurial touch and existing repository. '' '' Consult the [[#upgrade_Notes]] section to check how phase will move the first time a new version of Mercurial touch and existing repository.
Line 74: Line 74:
By default any changeset exchanged over the wire are set public. Advanced user may want a finer behavior. The Publishing repository concept is designed for this purpose. '' '' By default any changeset exchanged over the wire are set public. Advanced user may want a finer behavior. The Publishing repository concept is designed for this purpose.
Line 77: Line 77:
Setting a repository as "publishing" alter its behavior **when used as a server**: all changesets are **seen** as public changesets by clients. '' '' Setting a repository as "publishing" alter its behavior **when used as a server**: all changesets are **seen** as public changesets by clients.
Line 79: Line 79:
So, pushing to a "publishing" repository is the most common way to make changesets public: pushed changesets are seen as public on the remote side and marked as such on local side. '' '' So, pushing to a "publishing" repository is the most common way to make changesets public: pushed changesets are seen as public on the remote side and marked as such on local side.
Line 81: Line 81:
Note: the "publishing" property have no effects for local operations. '' '' Note: the "publishing" property have no effects for local operations.
Line 84: Line 84:
Phase is the first step of a series of features aiming at handling mutable history within mercurial. Old client do not support such feature and are unable to hold phase data. The safest solution is to consider as public any changeset going through an old client. '' '' Phase is the first step of a series of features aiming at handling mutable history within mercurial. Old client do not support such feature and are unable to hold phase data. The safest solution is to consider as public any changeset going through an old client.
Line 86: Line 86:
Moreover, most hosting solution will not support phase from the beginning. Having old clients seen as public repositories will not change their usage: public repositories where you push *immutable* public changesets *shared* with others. '' '' Moreover, most hosting solution will not support phase from the beginning. Having old clients seen as public repositories will not change their usage: public repositories where you push *immutable* public changesets *shared* with others.
Line 89: Line 89:
We discussed above that any changeset from a non-phase aware repository should be seen as public. This means that in the following scenario, X is pulled as public:: '' '' We discussed above that any changeset from a non-phase aware repository should be seen as public. This means that in the following scenario, X is pulled as public::
Line 97: Line 97:
We want to keep this behavior while creating/serving the A repository with new-hg. Although committing with any new-hg creates a draft changeset. To stay backward compatible, the pull must see the new commit as public. Non-publishing server will advertise them as draft. Having publishing repository the default is thus necessary to ensure this backward compatibility. '' '' We want to keep this behavior while creating/serving the A repository with new-hg. Although committing with any new-hg creates a draft changeset. To stay backward compatible, the pull must see the new commit as public. Non-publishing server will advertise them as draft. Having publishing repository the default is thus necessary to ensure this backward compatibility.
Line 99: Line 99:
This default value can also be expressed with the following sentence: "By default, without any configuration, everything you exchange with the outside is immutable.". This behaviour seems sane. '' '' This default value can also be expressed with the following sentence: "By default, without any configuration, everything you exchange with the outside is immutable.". This behaviour seems sane.
Line 102: Line 102:
Note: The publish option is aimed at controlling the behavior of ''server''. Changeset in any state on a publishing server will '''always''' be seen as public by other client. "Passive" repository which are only used as server for pull and push operation are not "affected" by this section. '' ''
Line 104: Line 103:
As in the choice for default, the main reason to allow draft changeset in publishing server is backward compatibility. With an old client, the following scenario is valid:: '' '' Note: The publish option is aimed at controlling the behavior of ''server''. Changeset in any state on a publishing server will '''always''' be seen as public by other client. "Passive" repository which are only used as server for pull and push operation are not "affected" by this section.

As in the choice for default, the main reason to allow draft changeset in publishing server is backward compatibility. With an old client, the following scenario is valid::
Line 114: Line 115:
If the default is publishing and new commits in such repository are "''public''" The following operation will be denied as X will be an '''immutable'' public changeset. However as other clients see X as public, any pull//push (or event pull//pull) will mark X as ''public'' in repo A. '' '' ''''' If the default is publishing and new commits in such repository are "''public''" The following operation will be denied as X will be an '''immutable'' public changeset. However as other clients see X as public, any pull//push (or event pull//pull) will mark X as ''public'' in repo A.
Line 116: Line 117:
'''''Allowing enforcement of public changeset only repository through config is probably something to do. This could be done with another "strict" option or a third value config for phase related option (mode=public, publishing(default), mutable) '' '' '''''
Line 119: Line 119:
----
''''' '' '' '' '' '''''
Line 122: Line 120:
'''''(If you were looking to the developer oriented page: PhaseDevel) '' '' ''''' The important point to are:

* repository with phase data can still accessed by old client
* new client will add phase data to any repository it touch
* If everything you plan to mutate is mq handled you don't have to care about
  anything.

=== Backward Compatility ===

Phase data are stored in a new files and does not alter any part of the existing
mercurial repository format. This means that a new client can safely write phase
related data without preventing an old client to works with the repository. This
allow new client store and handle phase related logic on **all repository**.

=== Adding phase data to old repo ===

There are a lot a repository out there with plenty of changeset but yet any
phase data. When looking at such repository, a new client take the safe road and
decided everything is 'public'. Some extension register logic to tune this
choice. For example, mq set every changeset under it's control as secret in this
situation.

You can set all changesets not pushed to a repository in the draft phase again
using:

{{{
hg phase --force --draft 'children(remote())'
}}}

=== Adding phase data to old repo ===

Beware that and old client won't be able to move phase when touching a repo.

* An old client can mutable immutable changesets

* An old client will push secret changeset

* An old client will commit new changeset in the phase of their parent

* An old client add changeset in the phase of their parent.

(If you were looking to the developer oriented page: PhaseDevel)

Phases

1. Introduction

The phase concept improves safety of history rewriting and control over changesets exchanged (read more). This phase concept aims to "just works" in a transparent manner for any user (read more). It is part of Core and always enabled in any new client but doesn't prevent older client to work on a repository (read more). Advanced user may decides to handle phase manually (read more).

Like bookmarks, phases are not stored in history and thus are not permanent and leave no audit trail.

This concept is introduced in Mercurial 2.1.

2. Available Phases

The phase concept allow to:

  • Prevent accidental rewriting part of the history expected to be immutable.
  • Keep immature changeset to be exchanged by mistake.

To achieve this, are three phases sharing a hierarchy of traits:

immutable

shared

public

x

x

draft

x

secret

* The public phases holds changeset announced publicly in. They are expect to

  • always exists in your history and are said immutable. History rewriting extension will refuse to delete such immutable changeset. Every changeset your push or pull from a public server are set in this phase.

* The draft phase holds changesets that was not expect marked as part of

  • the permanent history. You can safely rewrite them. New commit are draft by default.

* The secret phase holds changeset that you do not want to exchange with

  • other repository. Secret changeset are hidden to remote peer and won't be included in push operation. Manual operation or Extension may turn changeset secret.

Phases split the history in coherent set of changeset. Every changeset in a phase have ancestor in a phase compatible with its phase. Compatible means an changeset ancestors have at least the same traits that the children changeset. eg: A shared changeset alway have shared ancestor and an immutable changeset always have immutable ancestors.

In other word the phase of a changeset is alway equal of higher that the phase of it's descendant. According to the following order:

  • public < draft < secret

A changeset is not expected to automatically move from a lower phase to an higher phase (eg: from public to draft) but automatic

3. Phase Movements

Phase movement are automatic and transparent and most user don't have to care much about them. The base rule is very simple:

  • Any changesets on a remote repository is seen a public

On standard exchange commands, the phase of changesets on both side are compared. If differ the lowest phase is choosed. (eg: a changeset known as draft locally but public remotely is set public localy. Because public < draft)

This update happen during standard exchange commands:

* pull: remote phase data are used to update the phase data on

  • the local repo. As pull is read only, it does not change changeset's phase on the remote

* push: remote phase data are used to the phase data on the local repo and

  • then local phase data are pushed to the local repo.

The real behavior is a it's a bit more complicated than changesets on a remote repository is seen a public, but this is true for repository keeping default . If you need a finer behavior, consult the publishing repository section.

New changeset committed locally are draft, but some extension like mq may create secret and handle the move from secret to draft automatically

Consult the #upgrade_Notes section to check how phase will move the first time a new version of Mercurial touch and existing repository.

4. command line interface

4.1. core command

4.2. impact on extension

5. Publishing Repository

By default any changeset exchanged over the wire are set public. Advanced user may want a finer behavior. The Publishing repository concept is designed for this purpose.

5.0.1. What is a "publishing repository"?

Setting a repository as "publishing" alter its behavior **when used as a server**: all changesets are **seen** as public changesets by clients.

So, pushing to a "publishing" repository is the most common way to make changesets public: pushed changesets are seen as public on the remote side and marked as such on local side.

Note: the "publishing" property have no effects for local operations.

5.0.2. Old repository are publishing

Phase is the first step of a series of features aiming at handling mutable history within mercurial. Old client do not support such feature and are unable to hold phase data. The safest solution is to consider as public any changeset going through an old client.

Moreover, most hosting solution will not support phase from the beginning. Having old clients seen as public repositories will not change their usage: public repositories where you push *immutable* public changesets *shared* with others.

5.0.3. Why is "publishing" the default?

We discussed above that any changeset from a non-phase aware repository should be seen as public. This means that in the following scenario, X is pulled as public::

~/A$ cd ../B
~/B$ new-hg pull ../A # let's pretend A is served by old-hg
~/B$ new-hg log -r tip
summary: X phase: public

We want to keep this behavior while creating/serving the A repository with new-hg. Although committing with any new-hg creates a draft changeset. To stay backward compatible, the pull must see the new commit as public. Non-publishing server will advertise them as draft. Having publishing repository the default is thus necessary to ensure this backward compatibility.

This default value can also be expressed with the following sentence: "By default, without any configuration, everything you exchange with the outside is immutable.". This behaviour seems sane.

5.0.4. Why allow draft changeset in publishing repository

Note: The publish option is aimed at controlling the behavior of server. Changeset in any state on a publishing server will always be seen as public by other client. "Passive" repository which are only used as server for pull and push operation are not "affected" by this section.

As in the choice for default, the main reason to allow draft changeset in publishing server is backward compatibility. With an old client, the following scenario is valid::

~/A$ old-hg init
~/A$ echo 'babar' > jungle
~/A$ old-hg commit -mA 'X'
~/A$ old-hg qimport -r . # or any other mutable operation on X

If the default is publishing and new commits in such repository are "public" The following operation will be denied as X will be an immutable public changeset. However as other clients see X as public, any pull//push (or event pull//pull) will mark X as public in repo A.

6. Upgrade Notes

The important point to are:

* repository with phase data can still accessed by old client * new client will add phase data to any repository it touch * If everything you plan to mutate is mq handled you don't have to care about

  • anything.

6.1. Backward Compatility

Phase data are stored in a new files and does not alter any part of the existing mercurial repository format. This means that a new client can safely write phase related data without preventing an old client to works with the repository. This allow new client store and handle phase related logic on **all repository**.

6.2. Adding phase data to old repo

There are a lot a repository out there with plenty of changeset but yet any phase data. When looking at such repository, a new client take the safe road and decided everything is 'public'. Some extension register logic to tune this choice. For example, mq set every changeset under it's control as secret in this situation.

You can set all changesets not pushed to a repository in the draft phase again using:

hg phase --force --draft 'children(remote())'

6.3. Adding phase data to old repo

Beware that and old client won't be able to move phase when touching a repo.

* An old client can mutable immutable changesets

* An old client will push secret changeset

* An old client will commit new changeset in the phase of their parent

* An old client add changeset in the phase of their parent.

(If you were looking to the developer oriented page: PhaseDevel)

Phases (last edited 2014-12-04 14:50:53 by KimRandell)