Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Modeling » B3 » Branchpoint - literals or expressions?
Branchpoint - literals or expressions? [message #534972] Thu, 20 May 2010 16:40 Go to next message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
Hi,
currently branch points are literals (i.e. timestamp, revision, or tag).
This means they can not be set from properties (as this requires
evaluation). I think that should be possible.

I can still provide validation and code completion for the most common
case of using a literal string, and issue warnings that validation is
deferred until runtime if the expression is more complex.

Agree?

- henrik
Re: Branchpoint - literals or expressions? [message #534977 is a reply to message #534972] Thu, 20 May 2010 17:02 Go to previous messageGo to next message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
An alternative is naturally to allow property expansion in such
literals. The use of property expansion in strings have been avoided so
far. Don't know if they are required. They do facilitate "injection" of
properties that in turn have property expansion. If required, I would
prefer to do property expansion while loading properties (i.e. getting
properties from the command line, or from property files) as opposed to
having to perform property expansion on every string during evaluation.

- henrik

On 5/20/10 10:40 PM, Henrik Lindberg wrote:
> Hi,
> currently branch points are literals (i.e. timestamp, revision, or tag).
> This means they can not be set from properties (as this requires
> evaluation). I think that should be possible.
>
> I can still provide validation and code completion for the most common
> case of using a literal string, and issue warnings that validation is
> deferred until runtime if the expression is more complex.
>
> Agree?
>
> - henrik
Re: Branchpoint - literals or expressions? [message #535036 is a reply to message #534972] Fri, 21 May 2010 03:17 Go to previous messageGo to next message
Eike Stepper is currently offline Eike Stepper
Messages: 5499
Registered: July 2009
Senior Member
Hi Henrik,

Does a branch point *include* a reference to a branch?

Cheers
/Eike

----
http://thegordian.blogspot.com
http://twitter.com/eikestepper


Am 20.05.2010 22:40, schrieb Henrik Lindberg:
> Hi,
> currently branch points are literals (i.e. timestamp, revision, or
> tag). This means they can not be set from properties (as this requires
> evaluation). I think that should be possible.
>
> I can still provide validation and code completion for the most common
> case of using a literal string, and issue warnings that validation is
> deferred until runtime if the expression is more complex.
>
> Agree?
>
> - henrik
Re: Branchpoint - literals or expressions? [message #535085 is a reply to message #535036] Fri, 21 May 2010 05:00 Go to previous messageGo to next message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
The other way around, a repository has a (search) list of branches, a
branch has a branch point type, and branch point data.
E.g
main latest;
my_branch tag : "before_experiment3";

You may want to parameterize this so you can control the tag from the
command line, and write:
my_branch tag : $tag.experiment;

The branch point types are:
Latest
Tag
Revision
Timestamp

- henrik
Eike Stepper <stepper@esc-net.de> wrote:
> Hi Henrik,
>
> Does a branch point *include* a reference to a branch?
>
> Cheers
> /Eike
>
> ----
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
> > Hi,
> > currently branch points are literals (i.e. timestamp, revision, or >
> > tag). This means they can not be set from properties (as this
> > requires > evaluation). I think that should be possible.
> >
> > I can still provide validation and code completion for the most
> > common > case of using a literal string, and issue warnings that
> > validation is > deferred until runtime if the expression is more
> > complex.
> >
> > Agree?
> >
> > - henrik
>


--
- henrik
Re: Branchpoint - literals or expressions? [message #535105 is a reply to message #535085] Fri, 21 May 2010 05:36 Go to previous messageGo to next message
Eike Stepper is currently offline Eike Stepper
Messages: 5499
Registered: July 2009
Senior Member
Am 21.05.2010 11:00, schrieb Henrik Lindberg:
> The other way around, a repository has a (search) list of branches, a
> branch has a branch point type, and branch point data.
>
Do you mean something like:

branch ::= <branch name> <branch point type> [ ":" <branch point data>]

If so, I find it very confusing. IMHO it should be more like:

branch ::= <branch name>
<branch point> ::= <branch> <branch point type> [ ":" <branch point data>]

But I may not get the intent of all this correctly because in CDO we
also have branch and branch point concepts, where a branch point really
is a "point (in time) of a branch" ;-)

Cheers
/Eike

----
http://thegordian.blogspot.com
http://twitter.com/eikestepper



> E.g
> main latest;
> my_branch tag : "before_experiment3";
>
> You may want to parameterize this so you can control the tag from the
> command line, and write:
> my_branch tag : $tag.experiment;
>
> The branch point types are:
> Latest
> Tag
> Revision
> Timestamp
>
> - henrik
> Eike Stepper<stepper@esc-net.de> wrote:
>
>> Hi Henrik,
>>
>> Does a branch point *include* a reference to a branch?
>>
>> Cheers
>> /Eike
>>
>> ----
>> http://thegordian.blogspot.com
>> http://twitter.com/eikestepper
>>
>>
>> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
>>
>>> Hi,
>>> currently branch points are literals (i.e. timestamp, revision, or>
>>> tag). This means they can not be set from properties (as this
>>> requires> evaluation). I think that should be possible.
>>>
>>> I can still provide validation and code completion for the most
>>> common> case of using a literal string, and issue warnings that
>>> validation is> deferred until runtime if the expression is more
>>> complex.
>>>
>>> Agree?
>>>
>>> - henrik
>>>
>>
>
>
Re: Branchpoint - literals or expressions? [message #535113 is a reply to message #535105] Fri, 21 May 2010 06:29 Go to previous messageGo to next message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
Agree
In the current b3 model, the class is called Branch, it is really a
BranchPoint, because it has branch name, bp type and bp data as
features.
- henrik

Eike Stepper <stepper@esc-net.de> wrote:
> Am 21.05.2010 11:00, schrieb Henrik Lindberg:
> > The other way around, a repository has a (search) list of branches,
> > a
> > branch has a branch point type, and branch point data.
> > Do you mean something like:
>
> branch ::= <branch name> <branch point type> [ ":" <branch point
> data>]
>
> If so, I find it very confusing. IMHO it should be more like:
>
> branch ::= <branch name>
> <branch point> ::= <branch> <branch point type> [ ":" <branch point
> data>]
>
> But I may not get the intent of all this correctly because in CDO we
> also have branch and branch point concepts, where a branch point
> really is a "point (in time) of a branch" ;-)
>
> Cheers
> /Eike
>
> ----
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
>
> > E.g
> > main latest;
> > my_branch tag : "before_experiment3";
> >
> > You may want to parameterize this so you can control the tag from
> > the
> > command line, and write:
> > my_branch tag : $tag.experiment;
> >
> > The branch point types are:
> > Latest
> > Tag
> > Revision
> > Timestamp
> >
> > - henrik
> > Eike Stepper<stepper@esc-net.de> wrote:
> > >> Hi Henrik,
> >>
> >> Does a branch point *include* a reference to a branch?
> >>
> >> Cheers
> >> /Eike
> >>
> >> ----
> >> http://thegordian.blogspot.com
> >> http://twitter.com/eikestepper
> >>
> >>
> >> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
> >> >>> Hi,
> >>> currently branch points are literals (i.e. timestamp, revision,
> > > > or>
> >>> tag). This means they can not be set from properties (as this
> >>> requires> evaluation). I think that should be possible.
> >>>
> >>> I can still provide validation and code completion for the most
> >>> common> case of using a literal string, and issue warnings that
> >>> validation is> deferred until runtime if the expression is more
> >>> complex.
> >>>
> >>> Agree?
> >>>
> >>> - henrik
> >>> >> >
> >


--
- henrik
Re: Branchpoint - literals or expressions? [message #535119 is a reply to message #535113] Fri, 21 May 2010 06:59 Go to previous messageGo to next message
Thomas Hallgren is currently offline Thomas Hallgren
Messages: 3228
Registered: July 2009
Senior Member
So should we rename it to branch-points ?

- thomas

On 05/21/2010 12:29 PM, Henrik Lindberg wrote:
> Agree
> In the current b3 model, the class is called Branch, it is really a
> BranchPoint, because it has branch name, bp type and bp data as
> features.
> - henrik
>
> Eike Stepper<stepper@esc-net.de> wrote:
>> Am 21.05.2010 11:00, schrieb Henrik Lindberg:
>>> The other way around, a repository has a (search) list of branches,
>>> a
>>> branch has a branch point type, and branch point data.
>>> Do you mean something like:
>>
>> branch ::=<branch name> <branch point type> [ ":"<branch point
>> data>]
>>
>> If so, I find it very confusing. IMHO it should be more like:
>>
>> branch ::=<branch name>
>> <branch point> ::=<branch> <branch point type> [ ":"<branch point
>> data>]
>>
>> But I may not get the intent of all this correctly because in CDO we
>> also have branch and branch point concepts, where a branch point
>> really is a "point (in time) of a branch" ;-)
>>
>> Cheers
>> /Eike
>>
>> ----
>> http://thegordian.blogspot.com
>> http://twitter.com/eikestepper
>>
>>
>>
>>> E.g
>>> main latest;
>>> my_branch tag : "before_experiment3";
>>>
>>> You may want to parameterize this so you can control the tag from
>>> the
>>> command line, and write:
>>> my_branch tag : $tag.experiment;
>>>
>>> The branch point types are:
>>> Latest
>>> Tag
>>> Revision
>>> Timestamp
>>>
>>> - henrik
>>> Eike Stepper<stepper@esc-net.de> wrote:
>>> >> Hi Henrik,
>>>>
>>>> Does a branch point *include* a reference to a branch?
>>>>
>>>> Cheers
>>>> /Eike
>>>>
>>>> ----
>>>> http://thegordian.blogspot.com
>>>> http://twitter.com/eikestepper
>>>>
>>>>
>>>> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
>>>> >>> Hi,
>>>>> currently branch points are literals (i.e. timestamp, revision,
>>>>> or>
>>>>> tag). This means they can not be set from properties (as this
>>>>> requires> evaluation). I think that should be possible.
>>>>>
>>>>> I can still provide validation and code completion for the most
>>>>> common> case of using a literal string, and issue warnings that
>>>>> validation is> deferred until runtime if the expression is more
>>>>> complex.
>>>>>
>>>>> Agree?
>>>>>
>>>>> - henrik
>>>>> >> >
>>>
>
>
Re: Branchpoint - literals or expressions? [message #535135 is a reply to message #535113] Fri, 21 May 2010 08:21 Go to previous messageGo to next message
Eike Stepper is currently offline Eike Stepper
Messages: 5499
Registered: July 2009
Senior Member
Am 21.05.2010 12:29, schrieb Henrik Lindberg:
> Agree
> In the current b3 model, the class is called Branch, it is really a
> BranchPoint, because it has branch name, bp type and bp data as
> features.
>
Now I understand ;-)

Cheers
/Eike

----
http://thegordian.blogspot.com
http://twitter.com/eikestepper


> - henrik
>
> Eike Stepper<stepper@esc-net.de> wrote:
>
>> Am 21.05.2010 11:00, schrieb Henrik Lindberg:
>>
>>> The other way around, a repository has a (search) list of branches,
>>> a
>>> branch has a branch point type, and branch point data.
>>> Do you mean something like:
>>>
>> branch ::=<branch name> <branch point type> [ ":"<branch point
>> data>]
>>
>> If so, I find it very confusing. IMHO it should be more like:
>>
>> branch ::=<branch name>
>> <branch point> ::=<branch> <branch point type> [ ":"<branch point
>> data>]
>>
>> But I may not get the intent of all this correctly because in CDO we
>> also have branch and branch point concepts, where a branch point
>> really is a "point (in time) of a branch" ;-)
>>
>> Cheers
>> /Eike
>>
>> ----
>> http://thegordian.blogspot.com
>> http://twitter.com/eikestepper
>>
>>
>>
>>
>>> E.g
>>> main latest;
>>> my_branch tag : "before_experiment3";
>>>
>>> You may want to parameterize this so you can control the tag from
>>> the
>>> command line, and write:
>>> my_branch tag : $tag.experiment;
>>>
>>> The branch point types are:
>>> Latest
>>> Tag
>>> Revision
>>> Timestamp
>>>
>>> - henrik
>>> Eike Stepper<stepper@esc-net.de> wrote:
>>> >> Hi Henrik,
>>>
>>>> Does a branch point *include* a reference to a branch?
>>>>
>>>> Cheers
>>>> /Eike
>>>>
>>>> ----
>>>> http://thegordian.blogspot.com
>>>> http://twitter.com/eikestepper
>>>>
>>>>
>>>> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
>>>> >>> Hi,
>>>>
>>>>> currently branch points are literals (i.e. timestamp, revision,
>>>>> or>
>>>>> tag). This means they can not be set from properties (as this
>>>>> requires> evaluation). I think that should be possible.
>>>>>
>>>>> I can still provide validation and code completion for the most
>>>>> common> case of using a literal string, and issue warnings that
>>>>> validation is> deferred until runtime if the expression is more
>>>>> complex.
>>>>>
>>>>> Agree?
>>>>>
>>>>> - henrik
>>>>> >> >
>>>>>
>>>
>>>
>
>
Re: Branchpoint - literals or expressions? [message #535227 is a reply to message #535119] Fri, 21 May 2010 12:23 Go to previous messageGo to next message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
On 5/21/10 12:59 PM, Thomas Hallgren wrote:
> So should we rename it to branch-points ?
>
> - thomas
>
Nah, it works fine as we specify them in the order

branch name tag : foo

If we did

branch-point name tag : foo

it looks odd

Some other syntax would work better like:
pick tag : foo from branch name

or somesuch...

But I like to keep the syntax as it is.

We can rename it in the model if you think it makes a difference.

- henrik
Re: Branchpoint - literals or expressions? [message #535234 is a reply to message #535135] Fri, 21 May 2010 12:30 Go to previous message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
Good !
- henrik
On 5/21/10 2:21 PM, Eike Stepper wrote:
> Am 21.05.2010 12:29, schrieb Henrik Lindberg:
>> Agree
>> In the current b3 model, the class is called Branch, it is really a
>> BranchPoint, because it has branch name, bp type and bp data as
>> features.
> Now I understand ;-)
>
> Cheers
> /Eike
>
> ----
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
>> - henrik
>>
>> Eike Stepper<stepper@esc-net.de> wrote:
>>> Am 21.05.2010 11:00, schrieb Henrik Lindberg:
>>>> The other way around, a repository has a (search) list of branches,
>>>> a
>>>> branch has a branch point type, and branch point data.
>>>> Do you mean something like:
>>> branch ::=<branch name> <branch point type> [ ":"<branch point
>>> data>]
>>>
>>> If so, I find it very confusing. IMHO it should be more like:
>>>
>>> branch ::=<branch name>
>>> <branch point> ::=<branch> <branch point type> [ ":"<branch point
>>> data>]
>>>
>>> But I may not get the intent of all this correctly because in CDO we
>>> also have branch and branch point concepts, where a branch point
>>> really is a "point (in time) of a branch" ;-)
>>>
>>> Cheers
>>> /Eike
>>>
>>> ----
>>> http://thegordian.blogspot.com
>>> http://twitter.com/eikestepper
>>>
>>>
>>>
>>>> E.g
>>>> main latest;
>>>> my_branch tag : "before_experiment3";
>>>>
>>>> You may want to parameterize this so you can control the tag from
>>>> the
>>>> command line, and write:
>>>> my_branch tag : $tag.experiment;
>>>>
>>>> The branch point types are:
>>>> Latest
>>>> Tag
>>>> Revision
>>>> Timestamp
>>>>
>>>> - henrik
>>>> Eike Stepper<stepper@esc-net.de> wrote:
>>>> >> Hi Henrik,
>>>>> Does a branch point *include* a reference to a branch?
>>>>>
>>>>> Cheers
>>>>> /Eike
>>>>>
>>>>> ----
>>>>> http://thegordian.blogspot.com
>>>>> http://twitter.com/eikestepper
>>>>>
>>>>>
>>>>> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
>>>>> >>> Hi,
>>>>>> currently branch points are literals (i.e. timestamp, revision,
>>>>>> or>
>>>>>> tag). This means they can not be set from properties (as this
>>>>>> requires> evaluation). I think that should be possible.
>>>>>>
>>>>>> I can still provide validation and code completion for the most
>>>>>> common> case of using a literal string, and issue warnings that
>>>>>> validation is> deferred until runtime if the expression is more
>>>>>> complex.
>>>>>>
>>>>>> Agree?
>>>>>>
>>>>>> - henrik
>>>>>> >> >
>>>>
>>
Re: Branchpoint - literals or expressions? [message #608084 is a reply to message #534972] Thu, 20 May 2010 17:02 Go to previous message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
An alternative is naturally to allow property expansion in such
literals. The use of property expansion in strings have been avoided so
far. Don't know if they are required. They do facilitate "injection" of
properties that in turn have property expansion. If required, I would
prefer to do property expansion while loading properties (i.e. getting
properties from the command line, or from property files) as opposed to
having to perform property expansion on every string during evaluation.

- henrik

On 5/20/10 10:40 PM, Henrik Lindberg wrote:
> Hi,
> currently branch points are literals (i.e. timestamp, revision, or tag).
> This means they can not be set from properties (as this requires
> evaluation). I think that should be possible.
>
> I can still provide validation and code completion for the most common
> case of using a literal string, and issue warnings that validation is
> deferred until runtime if the expression is more complex.
>
> Agree?
>
> - henrik
Re: Branchpoint - literals or expressions? [message #608088 is a reply to message #534972] Fri, 21 May 2010 03:17 Go to previous message
Eike Stepper is currently offline Eike Stepper
Messages: 5499
Registered: July 2009
Senior Member
Hi Henrik,

Does a branch point *include* a reference to a branch?

Cheers
/Eike

----
http://thegordian.blogspot.com
http://twitter.com/eikestepper


Am 20.05.2010 22:40, schrieb Henrik Lindberg:
> Hi,
> currently branch points are literals (i.e. timestamp, revision, or
> tag). This means they can not be set from properties (as this requires
> evaluation). I think that should be possible.
>
> I can still provide validation and code completion for the most common
> case of using a literal string, and issue warnings that validation is
> deferred until runtime if the expression is more complex.
>
> Agree?
>
> - henrik
Re: Branchpoint - literals or expressions? [message #608089 is a reply to message #535036] Fri, 21 May 2010 05:00 Go to previous message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
The other way around, a repository has a (search) list of branches, a
branch has a branch point type, and branch point data.
E.g
main latest;
my_branch tag : "before_experiment3";

You may want to parameterize this so you can control the tag from the
command line, and write:
my_branch tag : $tag.experiment;

The branch point types are:
Latest
Tag
Revision
Timestamp

- henrik
Eike Stepper <stepper@esc-net.de> wrote:
> Hi Henrik,
>
> Does a branch point *include* a reference to a branch?
>
> Cheers
> /Eike
>
> ----
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
> > Hi,
> > currently branch points are literals (i.e. timestamp, revision, or >
> > tag). This means they can not be set from properties (as this
> > requires > evaluation). I think that should be possible.
> >
> > I can still provide validation and code completion for the most
> > common > case of using a literal string, and issue warnings that
> > validation is > deferred until runtime if the expression is more
> > complex.
> >
> > Agree?
> >
> > - henrik
>


--
- henrik
Re: Branchpoint - literals or expressions? [message #608092 is a reply to message #535085] Fri, 21 May 2010 05:36 Go to previous message
Eike Stepper is currently offline Eike Stepper
Messages: 5499
Registered: July 2009
Senior Member
Am 21.05.2010 11:00, schrieb Henrik Lindberg:
> The other way around, a repository has a (search) list of branches, a
> branch has a branch point type, and branch point data.
>
Do you mean something like:

branch ::= <branch name> <branch point type> [ ":" <branch point data>]

If so, I find it very confusing. IMHO it should be more like:

branch ::= <branch name>
<branch point> ::= <branch> <branch point type> [ ":" <branch point data>]

But I may not get the intent of all this correctly because in CDO we
also have branch and branch point concepts, where a branch point really
is a "point (in time) of a branch" ;-)

Cheers
/Eike

----
http://thegordian.blogspot.com
http://twitter.com/eikestepper



> E.g
> main latest;
> my_branch tag : "before_experiment3";
>
> You may want to parameterize this so you can control the tag from the
> command line, and write:
> my_branch tag : $tag.experiment;
>
> The branch point types are:
> Latest
> Tag
> Revision
> Timestamp
>
> - henrik
> Eike Stepper<stepper@esc-net.de> wrote:
>
>> Hi Henrik,
>>
>> Does a branch point *include* a reference to a branch?
>>
>> Cheers
>> /Eike
>>
>> ----
>> http://thegordian.blogspot.com
>> http://twitter.com/eikestepper
>>
>>
>> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
>>
>>> Hi,
>>> currently branch points are literals (i.e. timestamp, revision, or>
>>> tag). This means they can not be set from properties (as this
>>> requires> evaluation). I think that should be possible.
>>>
>>> I can still provide validation and code completion for the most
>>> common> case of using a literal string, and issue warnings that
>>> validation is> deferred until runtime if the expression is more
>>> complex.
>>>
>>> Agree?
>>>
>>> - henrik
>>>
>>
>
>
Re: Branchpoint - literals or expressions? [message #608093 is a reply to message #535105] Fri, 21 May 2010 06:29 Go to previous message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
Agree
In the current b3 model, the class is called Branch, it is really a
BranchPoint, because it has branch name, bp type and bp data as
features.
- henrik

Eike Stepper <stepper@esc-net.de> wrote:
> Am 21.05.2010 11:00, schrieb Henrik Lindberg:
> > The other way around, a repository has a (search) list of branches,
> > a
> > branch has a branch point type, and branch point data.
> > Do you mean something like:
>
> branch ::= <branch name> <branch point type> [ ":" <branch point
> data>]
>
> If so, I find it very confusing. IMHO it should be more like:
>
> branch ::= <branch name>
> <branch point> ::= <branch> <branch point type> [ ":" <branch point
> data>]
>
> But I may not get the intent of all this correctly because in CDO we
> also have branch and branch point concepts, where a branch point
> really is a "point (in time) of a branch" ;-)
>
> Cheers
> /Eike
>
> ----
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
>
> > E.g
> > main latest;
> > my_branch tag : "before_experiment3";
> >
> > You may want to parameterize this so you can control the tag from
> > the
> > command line, and write:
> > my_branch tag : $tag.experiment;
> >
> > The branch point types are:
> > Latest
> > Tag
> > Revision
> > Timestamp
> >
> > - henrik
> > Eike Stepper<stepper@esc-net.de> wrote:
> > >> Hi Henrik,
> >>
> >> Does a branch point *include* a reference to a branch?
> >>
> >> Cheers
> >> /Eike
> >>
> >> ----
> >> http://thegordian.blogspot.com
> >> http://twitter.com/eikestepper
> >>
> >>
> >> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
> >> >>> Hi,
> >>> currently branch points are literals (i.e. timestamp, revision,
> > > > or>
> >>> tag). This means they can not be set from properties (as this
> >>> requires> evaluation). I think that should be possible.
> >>>
> >>> I can still provide validation and code completion for the most
> >>> common> case of using a literal string, and issue warnings that
> >>> validation is> deferred until runtime if the expression is more
> >>> complex.
> >>>
> >>> Agree?
> >>>
> >>> - henrik
> >>> >> >
> >


--
- henrik
Re: Branchpoint - literals or expressions? [message #608095 is a reply to message #535113] Fri, 21 May 2010 06:59 Go to previous message
Thomas Hallgren is currently offline Thomas Hallgren
Messages: 3228
Registered: July 2009
Senior Member
So should we rename it to branch-points ?

- thomas

On 05/21/2010 12:29 PM, Henrik Lindberg wrote:
> Agree
> In the current b3 model, the class is called Branch, it is really a
> BranchPoint, because it has branch name, bp type and bp data as
> features.
> - henrik
>
> Eike Stepper<stepper@esc-net.de> wrote:
>> Am 21.05.2010 11:00, schrieb Henrik Lindberg:
>>> The other way around, a repository has a (search) list of branches,
>>> a
>>> branch has a branch point type, and branch point data.
>>> Do you mean something like:
>>
>> branch ::=<branch name> <branch point type> [ ":"<branch point
>> data>]
>>
>> If so, I find it very confusing. IMHO it should be more like:
>>
>> branch ::=<branch name>
>> <branch point> ::=<branch> <branch point type> [ ":"<branch point
>> data>]
>>
>> But I may not get the intent of all this correctly because in CDO we
>> also have branch and branch point concepts, where a branch point
>> really is a "point (in time) of a branch" ;-)
>>
>> Cheers
>> /Eike
>>
>> ----
>> http://thegordian.blogspot.com
>> http://twitter.com/eikestepper
>>
>>
>>
>>> E.g
>>> main latest;
>>> my_branch tag : "before_experiment3";
>>>
>>> You may want to parameterize this so you can control the tag from
>>> the
>>> command line, and write:
>>> my_branch tag : $tag.experiment;
>>>
>>> The branch point types are:
>>> Latest
>>> Tag
>>> Revision
>>> Timestamp
>>>
>>> - henrik
>>> Eike Stepper<stepper@esc-net.de> wrote:
>>> >> Hi Henrik,
>>>>
>>>> Does a branch point *include* a reference to a branch?
>>>>
>>>> Cheers
>>>> /Eike
>>>>
>>>> ----
>>>> http://thegordian.blogspot.com
>>>> http://twitter.com/eikestepper
>>>>
>>>>
>>>> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
>>>> >>> Hi,
>>>>> currently branch points are literals (i.e. timestamp, revision,
>>>>> or>
>>>>> tag). This means they can not be set from properties (as this
>>>>> requires> evaluation). I think that should be possible.
>>>>>
>>>>> I can still provide validation and code completion for the most
>>>>> common> case of using a literal string, and issue warnings that
>>>>> validation is> deferred until runtime if the expression is more
>>>>> complex.
>>>>>
>>>>> Agree?
>>>>>
>>>>> - henrik
>>>>> >> >
>>>
>
>
Re: Branchpoint - literals or expressions? [message #608097 is a reply to message #535113] Fri, 21 May 2010 08:21 Go to previous message
Eike Stepper is currently offline Eike Stepper
Messages: 5499
Registered: July 2009
Senior Member
Am 21.05.2010 12:29, schrieb Henrik Lindberg:
> Agree
> In the current b3 model, the class is called Branch, it is really a
> BranchPoint, because it has branch name, bp type and bp data as
> features.
>
Now I understand ;-)

Cheers
/Eike

----
http://thegordian.blogspot.com
http://twitter.com/eikestepper


> - henrik
>
> Eike Stepper<stepper@esc-net.de> wrote:
>
>> Am 21.05.2010 11:00, schrieb Henrik Lindberg:
>>
>>> The other way around, a repository has a (search) list of branches,
>>> a
>>> branch has a branch point type, and branch point data.
>>> Do you mean something like:
>>>
>> branch ::=<branch name> <branch point type> [ ":"<branch point
>> data>]
>>
>> If so, I find it very confusing. IMHO it should be more like:
>>
>> branch ::=<branch name>
>> <branch point> ::=<branch> <branch point type> [ ":"<branch point
>> data>]
>>
>> But I may not get the intent of all this correctly because in CDO we
>> also have branch and branch point concepts, where a branch point
>> really is a "point (in time) of a branch" ;-)
>>
>> Cheers
>> /Eike
>>
>> ----
>> http://thegordian.blogspot.com
>> http://twitter.com/eikestepper
>>
>>
>>
>>
>>> E.g
>>> main latest;
>>> my_branch tag : "before_experiment3";
>>>
>>> You may want to parameterize this so you can control the tag from
>>> the
>>> command line, and write:
>>> my_branch tag : $tag.experiment;
>>>
>>> The branch point types are:
>>> Latest
>>> Tag
>>> Revision
>>> Timestamp
>>>
>>> - henrik
>>> Eike Stepper<stepper@esc-net.de> wrote:
>>> >> Hi Henrik,
>>>
>>>> Does a branch point *include* a reference to a branch?
>>>>
>>>> Cheers
>>>> /Eike
>>>>
>>>> ----
>>>> http://thegordian.blogspot.com
>>>> http://twitter.com/eikestepper
>>>>
>>>>
>>>> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
>>>> >>> Hi,
>>>>
>>>>> currently branch points are literals (i.e. timestamp, revision,
>>>>> or>
>>>>> tag). This means they can not be set from properties (as this
>>>>> requires> evaluation). I think that should be possible.
>>>>>
>>>>> I can still provide validation and code completion for the most
>>>>> common> case of using a literal string, and issue warnings that
>>>>> validation is> deferred until runtime if the expression is more
>>>>> complex.
>>>>>
>>>>> Agree?
>>>>>
>>>>> - henrik
>>>>> >> >
>>>>>
>>>
>>>
>
>
Re: Branchpoint - literals or expressions? [message #608099 is a reply to message #535119] Fri, 21 May 2010 12:23 Go to previous message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
On 5/21/10 12:59 PM, Thomas Hallgren wrote:
> So should we rename it to branch-points ?
>
> - thomas
>
Nah, it works fine as we specify them in the order

branch name tag : foo

If we did

branch-point name tag : foo

it looks odd

Some other syntax would work better like:
pick tag : foo from branch name

or somesuch...

But I like to keep the syntax as it is.

We can rename it in the model if you think it makes a difference.

- henrik
Re: Branchpoint - literals or expressions? [message #608101 is a reply to message #535135] Fri, 21 May 2010 12:30 Go to previous message
Henrik Lindberg is currently offline Henrik Lindberg
Messages: 2499
Registered: July 2009
Senior Member
Good !
- henrik
On 5/21/10 2:21 PM, Eike Stepper wrote:
> Am 21.05.2010 12:29, schrieb Henrik Lindberg:
>> Agree
>> In the current b3 model, the class is called Branch, it is really a
>> BranchPoint, because it has branch name, bp type and bp data as
>> features.
> Now I understand ;-)
>
> Cheers
> /Eike
>
> ----
> http://thegordian.blogspot.com
> http://twitter.com/eikestepper
>
>
>> - henrik
>>
>> Eike Stepper<stepper@esc-net.de> wrote:
>>> Am 21.05.2010 11:00, schrieb Henrik Lindberg:
>>>> The other way around, a repository has a (search) list of branches,
>>>> a
>>>> branch has a branch point type, and branch point data.
>>>> Do you mean something like:
>>> branch ::=<branch name> <branch point type> [ ":"<branch point
>>> data>]
>>>
>>> If so, I find it very confusing. IMHO it should be more like:
>>>
>>> branch ::=<branch name>
>>> <branch point> ::=<branch> <branch point type> [ ":"<branch point
>>> data>]
>>>
>>> But I may not get the intent of all this correctly because in CDO we
>>> also have branch and branch point concepts, where a branch point
>>> really is a "point (in time) of a branch" ;-)
>>>
>>> Cheers
>>> /Eike
>>>
>>> ----
>>> http://thegordian.blogspot.com
>>> http://twitter.com/eikestepper
>>>
>>>
>>>
>>>> E.g
>>>> main latest;
>>>> my_branch tag : "before_experiment3";
>>>>
>>>> You may want to parameterize this so you can control the tag from
>>>> the
>>>> command line, and write:
>>>> my_branch tag : $tag.experiment;
>>>>
>>>> The branch point types are:
>>>> Latest
>>>> Tag
>>>> Revision
>>>> Timestamp
>>>>
>>>> - henrik
>>>> Eike Stepper<stepper@esc-net.de> wrote:
>>>> >> Hi Henrik,
>>>>> Does a branch point *include* a reference to a branch?
>>>>>
>>>>> Cheers
>>>>> /Eike
>>>>>
>>>>> ----
>>>>> http://thegordian.blogspot.com
>>>>> http://twitter.com/eikestepper
>>>>>
>>>>>
>>>>> Am 20.05.2010 22:40, schrieb Henrik Lindberg:
>>>>> >>> Hi,
>>>>>> currently branch points are literals (i.e. timestamp, revision,
>>>>>> or>
>>>>>> tag). This means they can not be set from properties (as this
>>>>>> requires> evaluation). I think that should be possible.
>>>>>>
>>>>>> I can still provide validation and code completion for the most
>>>>>> common> case of using a literal string, and issue warnings that
>>>>>> validation is> deferred until runtime if the expression is more
>>>>>> complex.
>>>>>>
>>>>>> Agree?
>>>>>>
>>>>>> - henrik
>>>>>> >> >
>>>>
>>
Previous Topic:Branches syntax
Next Topic:new b3 language version checked in
Goto Forum:
  


Current Time: Thu Aug 21 16:12:15 EDT 2014

Powered by FUDForum. Page generated in 0.02196 seconds