June 21, 2012 § Leave a comment
We try to read every blog post that goes through Mathblogging.org. For the Weekly Picks, we collect posts in one category from last week to give you an impression of what the mathematical blogosphere has to offer. (Read this for more information on this change.)
Last week, we focused on “Pure Researchers“.
- What’s on my blackboard? has captured some snark planning.
- At Combinatorics and more, Gil Kalai congratulates Tim Gowers to his knighthood.
August 17, 2011 § Leave a comment
After the big update last week not a lot has happened. We had to clean up a few smaller issues but nothing out of the ordinary (and thankfully no down time). The next few weeks will be slow because we’re all traveling.
Different code, different problems
The old code only ever used the live feeds — they were downloaded, processed to generate the pages and then discarded. This had some major drawbacks, especially for the stats since most feeds are too short to represent a whole month of posts.
Now that we’re temporarily storing information, we have to deal with other issues. For example, wordpress users mostly have rss feeds which do not offer a timestamp for the last update. So if something changes in a post (say, somebody adds a PlanetMO tag), the feed does not easily reflect this and our database does not update because checking more than the timestamp is rather cpu heavy. There’s not much we can do except appeal to bloggers to use the atom protocol, but since we’re only keeping information for a month, we hope it’s not a huge issue. After all, most posts never see an update.
After last weeks huge update only four new feeds found us and were added today.
August 8, 2011 § 4 Comments
The long awaited update is finally online.
After a bit of work, even more debugging and a lot of testing we have uploaded the new code. Even though the appearance has not changed much, the underlying technology is now much more efficient and reliable and will allow us to add more features in the long run.
The key: more data storage
The key change was the internal storage of all your regularly updated feeds. This is now done properly and allows much more reliable performance than our old code.
But it also helps with our features. For example, the statistics have always depended on the current rss-file from each blog. Almost all the statistics were simply wrong because the feeds are too short to store the information we needed for, say, the monthly count. Now that the feeds are stored in our database we can give more accurate counts.
Other cosmetic changes
The few cosmetic changes happened, e.g., on the “By Date” views that got a small menu. Most prominent are the changes to PlanetTAG. The tag cloud now has more entries and the dominant terms (such as “mathematics”) have been filtered out. We hope it will be more useful that way even though tagging is very inconsistent across the mathematical blogosphere.
Database update coming up
With these changes the code is finally efficient enough again to take on a whopping 100+ new database entries — getting us well on our way to 500 entries!
Stay tuned for more later this week!
July 28, 2011 § Leave a comment
Are we there yet?
Our re-write is complete. We’re testing the our code whenever we find time, but if all goes well, the new code will go online this weekend (and we will, of course, post about this here). There are some minor bugs we still need to hunt down and some cosmetic changes to fix.
All in all, we’re very excited to see if the code behaves as expected once it’s live. If all goes according to plan, virtually nothing will change on the surface of mathblogging.org, but things will finally become smoother under the hood.
We’ll keep you updated!
July 21, 2011 § 1 Comment
We thought you wouldn’t mind if we updated you a little on our coding work.
Yesterday we had one of your almost-weekly meetings, working across the big pond to work on the code. It has been four weeks since we started working on this major update. As it turns out the better description would be “complete re-write”.
Since mathblogging.org is our fun project and we’re all doing mathematical research to make a living, we don’t have a lot of time to work on the code. We realize that this is sometimes an inconvenience for the community, especially those 100 blogs that have been stuck in our delayed database update. As we wrote before, the size of the database as brought our puny amateur code to the breaking point. We cannot update the database before we finish this re-write or else the site will become very unreliable. Again, we’re sorry for any inconvenience that is already apparent (such as the infamous “white boxes” in the “by type” view).
The good news is that we’ve had a sort of debugging breakthrough yesterday when we finally solved a few major problems, getting the date view and type view going again. Next week we’re hoping to finish the debugging. We will test for at least one week locally before we push the update. So if all goes well, the biggest update so far (hey, we might even drop the “beta” ;) ) will be finished this month.
April 26, 2011 § Leave a comment
We just added some new entries to our database– that’s 263 entries now.
So a quick shout out is in order.
- Geometric Delights
- Math Prize for Math Community
- Απλά Μαθηματικά (Simple Math)
- Theory Announcements (Theorynet, DMANet)
- Geometry Bulletin Board
- Mr Palomar
- Flowing Data
- Cut The Knot
- Jim Wilder
- Mathematical Imagery
- Eugene B. Dynkin Collection of Mathematics Interviews
- The Statistics Forum
- Angry Math
April 21, 2011 § Leave a comment
Slowly we’re getting back into a weekly rhythm… What’s new?
We just deployed an updated version of our code to the app engine. (Un)fortunately, there’s nothing visible — yet.
After last week’s quota problems, we think we have identified and fixed the latest bug. If you care, here’s what happened. For safety reasons we had started to keep sanitized copies of each entry in the memcache (for things like PlanetMO). But some bloggers write such long posts that this increase blew the 1MB limit of GAE’s memcache. This in turn collided with the real bug — we had forgotten to catch such an error and good old GAE just retried to download and process such feeds indefinitely… Not too surprisingly, this led to an overload in incoming bandwidth and CPU-time.
So the only visible change with this update is that the “problematic” blogs (among them Timothy Gowers’s) are back in our database!
On the bright side, we are almost ready to release a couple of new features. So stay tuned for our next update (or hunt our code to find these easter eggs and let us know in the comments if you like what you see!).