Re: Summary of store_url project and some questions before posting some patches.

From: Alex Rousskov <rousskov_at_measurement-factory.com>
Date: Sat, 06 Oct 2012 15:48:50 -0600

On 10/06/2012 09:24 AM, Eliezer Croitoru wrote:
> As I moved forward and managed to make store_url feature stable to pass
> all of my tests The next step is to state a summary of the feature.
>
> The goal of store_url feature is to give squid a way to handle
> De-duplication of objects.
>
> Code implementation is quite simple but since I was working with the old
> bzr revision 12317 ant now its about 30 revisions up I dont have a clue
> on what to do.
> for now I downloaded the new 2a revision 12349.
>
>
> How it was done in squid 2.7 is less relevant for squid 3 branch
> implementation and was reviewed in the other thread.
>
> We discussed earlier on "url" to "originalUrl" refactoring and it seems
> to me that this will make a big and unneeded modification to the code
> that is not needed to achieve the goal.

One reason to rename the old "url" field is so that we can double check
that you caught all usage cases. If you do not rename, the patch will
not show the cases you missed (if any). Any renaming changes for this
reason alone can be later dropped (after reviews and before the commit).

Another reason is to alert future developers that they are dealing with
a URL which may be different from the store URL. If this is a valid/good
reason, the renaming should stay.

> Also At every place the store_url is mentioned can be the indication of
> it while every other place a "url" or "canonicalUrl" is mentioned will
> be the mark of original url usage.

Yes, but those "every other places" will not be visible in the patch and
since "url" is a rather generic term some of them may be difficult to
find in Squid sources using a manual search.

HTH,

Alex.
Received on Sat Oct 06 2012 - 21:49:10 MDT

This archive was generated by hypermail 2.2.0 : Mon Oct 08 2012 - 12:00:03 MDT