Manifest copy constructor does not deeply copy individual section Attributes
Roger Riggs
Roger.Riggs at oracle.com
Tue Jan 22 20:00:17 UTC 2019
Hi Phillip,
Did you discover need for a deeper copy? If so, can you elaborate?
It may be that adding a new constructor would be cleaner.
I'd suggest clarifying the documentation of existing methods as opposed
to changing the spec and implementation and the possible risks
to clients.
Thanks, Roger
On 01/21/2019 05:50 PM, Lance Andersen wrote:
> Hi Philipp,
>
> This is going to take some further discussion/input as this has been documented for sometime as a shallow copy.
>
> If there is a consensus to move this forward, it will also require a CSR. I can help with that and sponsoring the updates, if the decision is to go ahead with the proposed change
>
> Best
> Lance
>> On Jan 19, 2019, at 2:34 PM, Philipp Kunz <philipp.kunz at paratix.ch> wrote:
>>
>> Hi again,
>> Just after having sent the previous mail I found Manifest::clone
>> explicitly says it creates a shallow copy which is true only to a
>> certain extent. It deeply copies main attributes as well as individual
>> sections map already now and only shallowly copies individual sections
>> attributes maps.
>> I don't know about the background of it or why clone should only do a
>> shallow copy but if clone should really create a shallow copy, it
>> should probably create an even more shallow copy. In any case I figure
>> some potential for clarification at least.
>> Probably mostly because I already began a patch in the previous
>> message, I continued and attached another patch for a deep copy
>> approach. There might still be some reason not to deeply copy manifests
>> which I don't intend to forestall.
>> Philipp
>>
>> On Sat, 2019-01-19 at 19:32 +0100, Philipp Kunz wrote:
>>> Hi,
>>>
>>> Creating a new manifest as a copy from an existing one using the copy
>>> constructor assigns the new manifest individual sections entries map
>>> the same values which are references to attributes as the original
>>> rather than copying them as well deeply resulting in two manifests
>>> pointing to the same attributes instances modifications to which
>>> always affect both manifests. See test and proposed fix in the
>>> attached patch.
>>>
>>> Regards,
>>> Philipp
>> <manifestcopyconstructorandclone.patch>
> <http://oracle.com/us/design/oracle-email-sig-198324.gif>
> <http://oracle.com/us/design/oracle-email-sig-198324.gif> <http://oracle.com/us/design/oracle-email-sig-198324.gif>
> <http://oracle.com/us/design/oracle-email-sig-198324.gif>Lance Andersen| Principal Member of Technical Staff | +1.781.442.2037
> Oracle Java Engineering
> 1 Network Drive
> Burlington, MA 01803
> Lance.Andersen at oracle.com <mailto:Lance.Andersen at oracle.com>
>
>
>
More information about the core-libs-dev
mailing list