Welcome Guest, Not a member yet? Register   Sign In
Flash Data causes IE to balk
#11

[eluser]bretticus[/eluser]
[quote author="jhabbley" date="1253072732"][quote author="bretticus" date="1253071410"]
Man, I wish I would have saved my code to try that out. I, interestingly enough, got different values for "ci_session=". Do you have the encryption key set by chance?
[/quote]

For the purposes of this example, I disabled encryption. (I wanted to verify that the contents of each of the cookies was the same)

Before I disabled it, the encrypted values for ci_session were all different. But that's more of an effect of the encryption method, I believe. (mycrypt uses a randomized encryption method)

I've since re-enabled encryption, and everything is still working as expected.[/quote]

Good to know. Thanks!

Hate to patch the core files, but again, this really seems to be a bug.

Edit: Did you not choose to extend the Session class for a reason? Or is that one of the ones that can't be extended (shouldn't be if in the core library file I would think.)
#12

[eluser]jhabbley[/eluser]
The Session class is extendable, so it could be done that way.

However, I also believe this is a bug, and I do plan on submitting a bug report (perhaps tomorrow, free-time permitting)

In my opinion, extending the core classes is great if you intend to add on to or change the existing behavior to better suit your needs.

However, when the intended functionality of a core class causes problems, I think it's more efficient to correct the behavior at the source, rather than to add the overhead of extending a class. (detecting the extended class, loading the extended class, etc) This is especially true of a core class that is used on nearly every page.

Of course, this is my interpretation, and I'm certainly open to discussion.
#13

[eluser]bretticus[/eluser]
[quote author="jhabbley" date="1253095777"]
However, when the intended functionality of a core class causes problems, I think it's more efficient to correct the behavior at the source, rather than to add the overhead of extending a class. (detecting the extended class, loading the extended class, etc) This is especially true of a core class that is used on nearly every page.

Of course, this is my interpretation, and I'm certainly open to discussion.[/quote]

Funny you should mention that. Just tonight I have been stepping through the core files. I needed to extend the Config and Loader classes to use different views files for a mobile site if the host header matches the mobile site address. Anyways, I digress...

It appears to me that the MY_ extension libs in application are checked each time anyway and unless extends causes overhead, there's really no overhead using that method (unless you cause it within your extended class Smile )

I prefer not to update the core files in case they don't fix a bug in the next release. Which means if I upgrade the core files (next release), I've got to patch the core files again.




Theme © iAndrew 2016 - Forum software by © MyBB