March 2003 | ||||||
Sun | Mon | Tue | Wed | Thu | Fri | Sat |
1 | ||||||
2 | 3 | 4 | 5 | 6 | 7 | 8 |
9 | 10 | 11 | 12 | 13 | 14 | 15 |
16 | 17 | 18 | 19 | 20 | 21 | 22 |
23 | 24 | 25 | 26 | 27 | 28 | 29 |
30 | 31 | |||||
Feb Apr |
Nice to Be Back in the Saddle Again
Wow... 19 hours at the keyboard today less time for a shower, walk for donuts and walk for chinese food. Its been like 6 months since I did that kind of grind and oh what a feeling it is. Tired, cranky but huge feeling of accomplishment. I just wrote an OPML loader to take in Chris Pirillo favorite RSS feeds and then re-built and re-indexed the database.
Night.
11:12:10 PM Google It! comment [] IM Me About This
Roogle :: Top Ten Queries Via RSS
Thanks to inspiration from Brent / Net News Wire, I made an RSS feed out of the top ten queries. I'm not really comfortable with its appearance but I'm working on it.
Note: This is very loosely tested, only looked at in Radio's Aggregator, and could break other tools. I'll qa more tomorrow.
Tomorrow the changes will be much more under the hood as I (hopefully) get the phrase searching working.
7:18:59 PM Google It! comment [] IM Me About This
Roogle :: Updates
Ok. A few more things:
- Fix that ultimate act of brain deadness -- letting null searches throw errors. Thanks to Matt Mower for pointing this out
- Fixed a bunch of database schema errors
- Rebuilt and reindexed the database
- Added the ability for to drill to all postings by an author from a search results list -- Example (click the + sign)
- Added the ability to view / add comments in the blog's RSS feed supports it
- Sped it up a bit by removing an unneeded class
- Wrote admin tools for myself (yup -- that means its real, when you need tools).
- Fixed a bug displaying the RSS feeds. Thanks John !
- Roogle is now #8 on Blogdex. Go figure.
- Added the title into the index
- Recognized that there is a bunch more I need to learn about RSS. Time to visit Mark!
- Set my indexing routines up so I shouldn't have to go offline again (database / indices are now swappable by me)
- Recognized an interesting issue with indexing via RSS feeds -- if the RSS feed uses the LINK element for the item it is linking to then there is no way for me to link directly to the post only to the posts target. For an example, see Puzzlepieces. Now I know other, smarter, more talented folks than I have traveled down the "RSS War" path and I'm not gonna do that. I'll just keep reading up on this and figure out a possible solution.
Thanks to everyone for all the nice comments so far. Muchly appreciated and has sustained me through 14 hours of non-stop coding today. I even got a posting from Natrak over in Norway (and yes, my friend, I tossed your blog into the next index update pass).
Note: the total # of items in the database has fluctuated due to the database changes and 1 rather silly mistake when coding fast. It should now be correct and shouldn't again fluctuate to a smaller level.
7:02:22 PM Google It! comment [] IM Me About This
New Roogle Feature: Search Metrics
If you go to the Searches tab, you can see the search metrics, top ten hits, # of indexed documents, etc. Not yet an RSS feed (I have it working but the feed doesn't validate for reasons that escape me). [_Go_]
1:03:52 PM Google It! comment [] IM Me About This
New Feature: Adding Metadata to Search Results
Ok. Here's a cool (well I think its cool at least -- let me know if its useless) new thing. What I've done is added metadata to search results so you can look at a search result and see visually whether that post has:
- Multiple Links
- Any images
- Any mailto links
Why? What Was I Thinking?
We all surf the net a lot and while we might not be able to remember an exact posting, we tend to be able to remember some of its characteristics -- WHEN they deviate from the norm AND if they're visual in nature. When I think about cool cell phones for example, I'd naturally think of Russell. But Russell has written so much on cell phones that if the search result can tell me "Has 5 pictures" then I can get a rough feel if this might be the posting. Similarly if I'm looking for a post where I remember that it had a mailto link then I can now tell from the search results. Now both mailtos and pictures are still somewhat uncommon in blog posts so I start the count on those from 1 (i.e. I'll display their count if there is at least 1). But we assume that blog posts have at least 1 url always so I only show urls when there are more than 1 (which might not be the right thresh hold / is 5 better).
Sound silly? Here are some examples:
Should It Stay or Should It Go Now?
Let me know if this is worthwhile or simply featuritis and should be eliminated (and yes that means that The Clash is playing now on my iPod).
11:27:31 AM Google It! comment [] IM Me About This
Roogle Early Morning Update
I woke early with sugarplums dancing in my head ... Wait? I missed Christmas? Oh that's right I woke up with visions of Roogle metadata (# of queries, top ten queries, etc) in my head. This is a great suggestion from Brent at Net News Wire and will be done later this morning. Hm... Since I have my iBook again, I'm thinking I should download it and give it a try. Everyone seems to love it.
What's new right is self service adding of your own blog! Just go here and fill out the form and you're in. Don't worry about whether or not you're in there already -- each RSS url is validated to only exist ONCE in the database and the blog url and RSS url are validated to actually exist on the net as readable URLs before they are accepted.
Note: If you emailed me about being added, no need to fill out the form again, I'll take care of it.
7:14:32 AM Google It! comment [] IM Me About This