2009/7/3 Karsten Wade <kwade(a)redhat.com>:
On Thu, Jul 02, 2009 at 11:19:38PM +0200, Yaakov Nemoy wrote:
> Hey List,
>
> Max and I had a frank discussion and we came to the conclusion that we
> needed a quick hack to get results.
[snip]
> This release is also going to include a bunch of coding fixes but it's
> not perfect. Any statistics generated with this release should be
> taken with a grain of salt. A very big one.
These statements seem to contradict, if we cannot trust the results
that we quickly hacked to get.
Were these coding fixes necessary? By comparison, I can take an HTML
table, copy/paste from the browser in to oocalc, fudge around with the
results, and get a CSV fairly quickly. I'd rather do that and trust
the statistics, than get a nice CSV that is only good for rubbing the
rim of a margarita glass.
Let me be a bit clearer about what kind of errors were encountering.
We're trying to pull data that's been encoded in at least one codec,
sometimes without being totally correctly labelled. This can include
one instance where there is one encoding for the email, a second
encoding for the name, a third encoding for the website presented by
RH as an archive, and escape sequences throughout to manage all this.
The kinds of statistics that get short changed by this are ones where
people use things like accents in their name, and such. This would
affect both our HTML output and our CSV output, since it really
affects our ability to scan hundereds of thousands of webpages without
the entire program crashing and burning.
-Yaakov