Re: [Semanticscuttle-devel] Privatekey

[Mark P. <mpe...@gm...>]

Sorry for the delay on the privatekey code.  I'm finishing up with the tests
and should be done shortly.  Cheers to all.
Mark


On Mon, Mar 7, 2011 at 4:08 PM, Mark Pemberton <mpe...@gm...>wrote:

> Thanks for the great response Christian!  I'll work on these items and get
> back to you.  Cheers to all!
>
> On Mon, Mar 7, 2011 at 2:03 PM, Christian Weiske <cw...@cw...>wrote:
>
>> Hi Mark,
>>
>>
>>
>> > Christian, I've updated privatekey with the following changes:
>> > 1) I added an enable checkbox to allow user to enable/disable the
>> > Private RSS feed.
>> That's a good addition since it allows people to easily deactivate
>> their private key without manually deleting the key.
>>
>> I saw that you added a second column beside the private key;
>> enablePrivateKey. In my eyes this is not necessary since that
>> information can be transmitted in the private key column - when it's
>> empty/NULL, it's deactivated. Please change that.
>>
>> > 2) I took out the ajax call in profile.php - I ran into some issues
>> > with trying keep the javascript for those who use it.  Didn't think
>> > it was that big enough of a deal to worry about it.
>> Fine with me.
>>
>> I've got a little special configuration on my dev system:
>> > $sitename = 'b"m.bo\'go';
>> The site name here contains both single and double quotes, something
>> that helped me discovering quite a number of quoting/escaping bugs. It
>> seems that you double-htmlencode the title in the rss feed so that I
>> get
>> > title="b&amp;quot;m.bo'go: (private) b&amp;quot;m.bo'go"
>> instead of the
>> > title="b&quot;m.bo'go: Recent bookmarks"
>> as it is for the normal feed in the HTML head. Also add a newline
>> before the private <link> so the html looks cleaner. You also do not
>> escape the & in the link before privatekey=, which makes the page fail
>> to validate.
>>
>> There is no fail message when an unknown privatekey is used.
>> SemanticScuttle should send a status 400 (Bad request) with an
>> explanatory message.
>>
>> The user's bookmark list does also not contain a privatekey feed link.
>> I think that all pages that link feeds should also link the
>> associated private feed (i.e. useful when someone wants to bookmark a
>> search that may contain private bookmarks).
>>
>> And unit tests are a must here:
>> - fetch feed with private key, do the private bookmarks appear?
>> - same without priv key
>> - are the private feed links on all relevant pages when logged in and
>>  the user has a private key?
>> - same for logged out -> no private links
>> - same for logged in bug no private key -> no private feed links
>>
>>
>> > 3) on a separate commit, I took a stab at updating the getBookmarks()
>> > function to make the queries a little more efficient.  The base of the
>> > problem was doing both a GROUP BY and an ORDER BY in the same query
>> > (as well as the use of multiple tables on the same query).  I got
>> > spoiled on Oracle because it tends to be a little more efficient.
>> > MySQL as well as other RDBS's don't optimize it enough to make it
>> > very efficient.  So I tweaked the queries to get around some of the
>> > issues.  I haven't tested it on MySQL 4 so I can't confirm if the
>> > query will work on that version.  There was one component that I
>> > couldn't make more efficient and that is searching by a tag, or by a
>> > user, or searching.  Those queries are simply too complex to make
>> > efficient.  However, they are fast if there is only a small
>> > resultset. Anyway, outside of doing a full redesign of the data
>> > tables, this is the best I could come up with.  My last commit
>> > contains the updates to the user table and the bookmarks table.  I
>> > simply added an additional index on the bookmarks table to help with
>> > the query.  Please let me know if you have any questions.
>>
>> Do the tests do still run? I get a bunch of errors when running them.
>> Please also do that in a different branch; we're using one branch per
>> feature which makes it easier to merge the relevant features in the
>> master branch later. Just imagine that the private key feature is
>> ready, but your sql optimizations are buggy but we want to release a
>> new version. If we have clean feature branches, we just merge the ones
>> that are ready and are done. Currently this is not possible since an
>> unfinished feature is in the same branch as the other one.
>>
>> --
>> Regards/Mit freundlichen Grüßen
>> Christian Weiske
>>
>> -=≡ Geeking around in the name of science since 1982 ≡=-
>>
>>
>> ------------------------------------------------------------------------------
>> What You Don't Know About Data Connectivity CAN Hurt You
>> This paper provides an overview of data connectivity, details
>> its effect on application quality, and explores various alternative
>> solutions. http://p.sf.net/sfu/progress-d2d
>>
>> _______________________________________________
>> Semanticscuttle-devel mailing list
>> Sem...@li...
>> https://lists.sourceforge.net/lists/listinfo/semanticscuttle-devel
>>
>>
>