Re: copy newly added file

[Fejes József <fe...@joco.name>]

>>>> I know that a newly added file doesn't have a history yet but why  
>>>> does
>>>> that matter? That + sign could appear just like it does with another
>>>> file that does have a history, it should be a link nonetheless. I  
>>>> don't
>>>
>>> Well the + means "copied with history", but clearly from svn's
>>> perspective there is no history, so I think the output is correct. I
>>> don't know whether the developers think that the overall behavior is
>>> correct.
>>
>> The lack of + made me think that these files are not cheap copies,  
>> just
>> regular, space-wasting copies, but clearly I was wrong about the  
>> meaning
>> of it. So is it cheap or not? How do I tell the difference without  
>> the +
>> sign?
> 
> I don't know if this copy will be cheap. But you can create a  
> throwaway repository and try it out.
> 
> If you find the copy is not cheap, and you don't want to use file- 
> level externals for this, then I recommend you do the other thing you  
> don't want to do, and commit the lib directory first, then create the  
> web* directories and copy things to them and commit them second. You  
> say it's a pain, but if this is a type of build you do often, I  
> expect you already have a script for it, so just change these steps  
> in the script.

Ok, I experimented a little with a local repository, using the latest
version of TortoiseSVN. I found that the copy of a newly added file is not
cheap. Should I report this as a bug then?

And thanks for pointing out file level externals. The idea sounds better
than copying, I'll stick to that. But let's not close the above issue yet
about cheap copies.

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2379519

To unsubscribe from this discussion, e-mail: [users-unsubscribe@subversion.tigris.org].

Re: copy newly added file

[Andy Levy <an...@gmail.com>]

2009/8/3 Fejes József <fe...@joco.name>:
>>> Ok, I experimented a little with a local repository, using the latest
>>> version of TortoiseSVN. I found that the copy of a newly added file is
>>> not
>>> cheap. Should I report this as a bug then?
>>
>> Cheap copies require that there be something already in the repository
>> to reference (the copy is just a pointer to the source @ a particular
>> revision). If the item being copied isn't already in the repository, I
>> don't understand how the copy could be made cheaply.
>
> I don't understand either, but it's not my job to :) So you think it's
> rather a feature request than a defect? I know that a cheap copy is a
> pointer, but it might as well point to the same revision, not just a
> previous one, right? Or maybe it's just not worth it.
>

Definitely feels like a feature request to me, based upon my
understanding things are working as designed here.

It may not be possible to implement with the current design of SVN, as
to make the copy you need to know the revision number you're copying
from, and that doesn't exist yet in the scenario you're describing.

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2379526

To unsubscribe from this discussion, e-mail: [users-unsubscribe@subversion.tigris.org].

Re: copy newly added file

[Fejes József <fe...@joco.name>]

>> Ok, I experimented a little with a local repository, using the latest
>> version of TortoiseSVN. I found that the copy of a newly added file is
>> not
>> cheap. Should I report this as a bug then?
> 
> Cheap copies require that there be something already in the repository
> to reference (the copy is just a pointer to the source @ a particular
> revision). If the item being copied isn't already in the repository, I
> don't understand how the copy could be made cheaply.

I don't understand either, but it's not my job to :) So you think it's
rather a feature request than a defect? I know that a cheap copy is a
pointer, but it might as well point to the same revision, not just a
previous one, right? Or maybe it's just not worth it.

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2379523

To unsubscribe from this discussion, e-mail: [users-unsubscribe@subversion.tigris.org].

Re: copy newly added file

[Andy Levy <an...@gmail.com>]

2009/8/3 Fejes József <fe...@joco.name>:
>>>>> I know that a newly added file doesn't have a history yet but why
>>>>> does
>>>>> that matter? That + sign could appear just like it does with another
>>>>> file that does have a history, it should be a link nonetheless. I
>>>>> don't
>>>>
>>>> Well the + means "copied with history", but clearly from svn's
>>>> perspective there is no history, so I think the output is correct. I
>>>> don't know whether the developers think that the overall behavior is
>>>> correct.
>>>
>>> The lack of + made me think that these files are not cheap copies,
>>> just
>>> regular, space-wasting copies, but clearly I was wrong about the
>>> meaning
>>> of it. So is it cheap or not? How do I tell the difference without
>>> the +
>>> sign?
>>
>> I don't know if this copy will be cheap. But you can create a
>> throwaway repository and try it out.
>>
>> If you find the copy is not cheap, and you don't want to use file-
>> level externals for this, then I recommend you do the other thing you
>> don't want to do, and commit the lib directory first, then create the
>> web* directories and copy things to them and commit them second. You
>> say it's a pain, but if this is a type of build you do often, I
>> expect you already have a script for it, so just change these steps
>> in the script.
>
> Ok, I experimented a little with a local repository, using the latest
> version of TortoiseSVN. I found that the copy of a newly added file is not
> cheap. Should I report this as a bug then?

Cheap copies require that there be something already in the repository
to reference (the copy is just a pointer to the source @ a particular
revision). If the item being copied isn't already in the repository, I
don't understand how the copy could be made cheaply.

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2379522

To unsubscribe from this discussion, e-mail: [users-unsubscribe@subversion.tigris.org].