Feedster Debugging Humor

I know I changed some things yesterday but I'm right now at a position where all I can tell you is "damificanremember".  So instead here's an interesting and hopefully humorous debugging anecdote.  I got an IM last night from someone who's been very helpful to me with respect to Feedster.  And his comment was that we're not indexing him anymore since he can't find recent urls he's posted in our url index.  That was a feature we just added so of course its suspect.  Now this started a debugging trail like this:

• Me: Well it could definitely be my fault (I don't have a lot of pride when it comes to bugs).  What's the url?
• Him: http WHATEVER
• Me: (Thinking to myself -- this doesn't look like his blog but its so damn hard to keep the relationship between IM name, email name, blog url, etc, so I'll just check the RSS).
• Me: There aren't any hyperlinks in the RSS url?
• Him: Yes there are.
• Me: No there aren't and give him back http WHATEVER / rss
• Him: That's not me -- that's what I was linking to.
• Me: oh.
• Me: What's your RSS url?
• Him: httpWHATEVER
• Me: Look at that
• Me: Cast chicken bones and sacrifice a goat and dance next to a roaring fire while chanting and taking into account factors like:
• Are there really new entries in the RSS?  yes.
• Force the indexing process for just that url.  no difference.
• Ask if he updates weblogs.com when he updates?  yes.
• Think about my recent switch to etags for only getting recently modified rss feeds.  no difference.
• Run other scenarios in my head mostly involving any number of dumb things I could have done.
• Look at the rss feed and notice that the new urls in the feed aren't in the back end database
• Me: Everything seems right ...  But I keep getting "stale" objects for the rss feed like I can't read it.Him:
• Him: Well I've been having a routing problem.  Could you trace route to me?
• Me: Pray to the gods of IP networking and rub the statue of Vinton Cerf I keep on my desk and then execute the magical Trace Route command (of course its not in the path).  And ... site can't be found.  Site can't be pung. 
• Him: Can I have the NSLooup records?
• Me: Here ya go!  TTYL (talk to you later in IM speak)
• Him: OK, time to flog my  Unix admin. (that's a quote!)

Morals of the story:

1. Its not always your (or my) fault.
2. Always, always, always look at the input data.  Even if I had been able to get to the rss feed, there would have been another problem that would have prevented it from working.
3. Don't disregard the fact that it could be your fault.  Making errors is **damn easy**.  Particularly with something like a search engine.
4. With network applications like a search engine what you see at your desktop (his site came up) DOESN'T MEAN SQUAT!  All that matters is what your server sees.  Its a good argument for using your server's DNS settings not your ISPs.

10:37:08 AM      Google It!   comment []    IM Me About This   

Rant: Comments on Running a Conference from a Speaker's Perspective

<RANT>

Now I know damn little about running a conference but I know a lot about being a speaker -- I've spoken at I don't know how many conferences ranging from trade shows to academic forums and some things remain the same:

• Presentations are written at the last minute
• Presentations are written with the tools people know UNLESS a toolset is specified in advance

Now I don't want to throw stones and name names but an upcoming conference and a conference that by all rights I should be heavily in favor of is breaking both these rules:

• They're now down to less than 2 months before the conference starts and speakers WHO SUBMITTED ABSTRACTS on time still aren't notified of acceptance or non-acceptance despite slots being open on the schedule.  People that did submit abstracts now feel like "oh they'll take me but only if nothing better comes in at the last minute".  So what is almost certain to happen is the remaining speakers will get notified at the last minute and then they'll be writing them right at the end.
• They've post facto "recommended" that slides should really be marked up in some new XML format.  Now maybe this is a good thing -- but its not what anyone signed on for.  If you want them that way then decide in advance.  Don't change the rules afterwards.

</RANT>

10:07:03 AM      Google It!   comment []    IM Me About This