On 9/5/2012 8:45 AM, Ramchandra Apte wrote:
On Wednesday, 5 September 2012 15:03:56 UTC+5:30, Terry Reedy wrote:
On 9/5/2012 1:22 AM, Ramchandra Apte wrote:
On Wednesday, 5 September 2012 09:35:43 UTC+5:30, Terry Reedy wrote:
If you do find the right place, you should contribute something to an
improvement. The current search performance is not a secret, so mere
complaints are useless.
These ever increasing extra blank lines with each quote are obnoxious.
Consider using a news reader with news.gmane.org instead of google crap.
Or snip heavily.
Google site search costs 2000$ for 500,000 searches per year and 750$ for
150,000 searches so its quite expensive.
Also the print function only comes in the third result (python 3.2)
if you search for "site:docs.python.org/release/3.2 print" the print function
is not found at all.
I think a specialized algorithm would work better.
I'm going to code an program for this.
A simple algorithm would be to present index search results first, if
there are any, and then page search results.
Then searching print would return
"Index entries for print:"
Builtin-functions page
a couple of others...
Pages containing print:
<list of about 150 pages>
I would not worry about duplication.
Labeling index results as such would clue people in to the fact that
they could have looked for the object name in the index. People names
like 'Lundh' that are not indexed but which appear on several pages
would give the same result as before.
Looking at the web page (which I do not normally use), I see that the
problem is deeper. The left margin of every page have an inviting "Quick
search" box with text "Enter search terms or a module, class or function
name." But it does not currently work very well for such object names.
The index is only available from the main contents page.
This contrasts with the Windows docs which has an index tab, making the
index directly available from *anywhere*. (There is also a separate text
search tab.) I think an index search box should be added above the text
search box. I will ask on pydev where the suggestion should go.
--
Terry Jan Reedy
--
http://mail.python.org/mailman/listinfo/python-list
