Re: raw rows post-processing, (im)mutable?
Re: raw rows post-processing, (im)mutable?
- Subject: Re: raw rows post-processing, (im)mutable?
- From: Chuck Hill <email@hidden>
- Date: Wed, 24 Feb 2016 00:11:49 +0000
- Thread-topic: raw rows post-processing, (im)mutable?
In theory your more optimized code is relying on an undocumented implementation detail which is bad. The next release of WO could break your code. But as Donald Trump has a better chance of being selected as the next Pope than of Apple making a new WO release it seems safe in actual practice.
You could do something like this:
if (needToAddSomething(rr))
def nrr= rr instanceof NSMutableDictionary ? rr : rr.mutableClone()
On 2016-02-23, 4:04 PM, "webobjects-dev-bounces+chill=email@hidden on behalf of OC" <webobjects-dev-bounces+chill=email@hidden on behalf of email@hidden> wrote:
>Hello there,
>
>well, having fetched raw rows all right, I need to post-process them, essentially by checking each fetched raw row, and to some of them (not all, but many) adding couple of extra attributes.
>
>The safe and clean approach would, of course, be something like
>
>=== safe&clean ===
>def fetchedRawRows=ec.objectsWithFetchSpecification(fs)
>def result=new NSMutableArray()
>for (rr in fetchedRawRows)
> if (needToAddSomething(rr))
> def nrr=rr.mutableClone()
> nrr.setObjectForKey(...) // more of these
> nrr.removeObjectForKey(...) // when post-processed, some fetched BLOBs are not needed anymore, NSData will be GCed in due time
> result.addObject(nrr)
> } else result.addObject(rr)
>return result
>===
>
>in practice though those returned raw rows seem to be mutable (com.webobjects.eocontrol._EOMutableKnownKeyDictionaries, whatever that means), and the unsafe unclean but considerably more efficient approach actually seems to work:
>
>=== quick&dirty ===
>def fetchedRawRows=ec.objectsWithFetchSpecification(fs)
>for (rr in fetchedRawRows)
> if (needToAddSomething(rr)) {
> rr.setObjectForKey(...) // more of these
> rr.removeObjectForKey(...) // unlike above, might be GCed immediately if need be
> }
>return fetchedRawRows
>===
>
>Since fetchedRawRows will tend to be rather at the huge side, quite probably the difference will be notable.
>
>Can someone say by experience or by a deep knowledge that the quick&dirty approach is actually safe? Or, contrariwise, that if I embrace it, sooner or later the hell will break loose?
>
>Thanks!
>OC
>
>
> _______________________________________________
>Do not post admin requests to the list. They will be ignored.
>Webobjects-dev mailing list (email@hidden)
>Help/Unsubscribe/Update your Subscription:
>
>This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden