Cal Henderson on why he hates Django, covers issues of scaling and how Django may implement better scaling features in the future. Some of these are likely implemented now.
Guido on the Google App Engine:
Thursday, October 21, 2010
Sunday, October 3, 2010
upgrading PostgreSQL on Debian/Ubuntu
I've been upgrading pg on a few servers today. One from 8.1 to 8.4 (debian) and the other from 8.2 to 8.4 (ubuntu). In both cases the process worked as noted here (though I used aptitude):
Once you're sure that's working:
- backup your db(s)
- add backports repositories ubuntu, debian instructions (don't forget about pinning)
aptitude install postgresql-8.4
pg_dropcluster --stop 8.4 main
pg_upgradecluster 8.1 main
Once you're sure that's working:
pg_dropcluster --stop 8.1 main
apt-get remove postgresql-8.1
Friday, October 1, 2010
mysql string search and replace
UPDATE [tablename] SET [fieldname] = REPLACE([fieldname],"[oldstring]","[newstring]");
The code above would operate on every row in the table but you may want WHERE conditions. Example:
UPDATE profile SET homepage = REPLACE(homepage,"blargh.example.com","foo.example.com") WHERE first_name="Joe" AND last_name="Schmoe";
Will change Joe's homepage to http://foo.example.com/ instead of http://blargh.example.com/.
Sunday, August 15, 2010
Sunday, August 8, 2010
more altruistic engineers
Awesome story about the devs who created "Graphing Calculator 1.0, which Apple bundled with the original PowerPC computers". They basically built it on their own time, unofficially, at Apple without being Apple employees and without official approval.
Saturday, August 7, 2010
great video on the Python GIL
Taught me more than I wanted to know about how the GIL works and reveals a critical problem.
Friday, August 6, 2010
reveals a lot about these communities
More than 1 in 10 Mozilla bug finders turn down cash. Both that mozilla pays for finding bugs and that a number of devs turn down the cash.
Tuesday, March 2, 2010
Wednesday, February 24, 2010
db design at reddit
There's an interesting post in this programming reddit thread. The user "spez" was one of the reddit sysadmins for a number of years. I'm not sure if he still works there or not.
Tuesday, February 16, 2010
is xhtml really the way to go?
For a while now I've been using XHTML for a number of pages but this post from 2006 from the webkit blog made me rethink. The gist of the post is that because IE sucks you have to do crazy things with content type headers to truly support XHTML so you may as well serve HTML4. I'm not sure I'm ready to change yet but the argument is compelling.
More info here in this reddit programming thread as well.
More info here in this reddit programming thread as well.
migrating a facebook FBML app to IFRAME
I've been migrating a facebook app to iframe from FBML so I'll be posting about that process a bit here. I was figuratively beating my head against a wall trying to determine why some of my xfbml wasn't working. After much ado I came across this thread which posed the question "I can connect to XFBML successfully if I create a login button, that when clicked on will connect to facebook connect and log on. However why do I actually need to click on this login button?" This was exactly my problem though apparently I couldn't phrase my question appropriately for google.
So the answer is, if it doesn't seem like your xfbml or your serverfbml code is working properly (but it is sorta working) you probably don't have your facebook connect app settings setup correctly. Particularly the "base domain" and/or connect url.
I know you're not creating a facebook connect site at all so you'd think these settings were irrelevant. You, like me, would be wrong.
So the answer is, if it doesn't seem like your xfbml or your serverfbml code is working properly (but it is sorta working) you probably don't have your facebook connect app settings setup correctly. Particularly the "base domain" and/or connect url.
I know you're not creating a facebook connect site at all so you'd think these settings were irrelevant. You, like me, would be wrong.
Subscribe to:
Posts (Atom)