…and this is a 
Technical difficulties!
Category: Administrative
Posted on: May 11, 2007 1:15 PM, by PZ Myers
The server here is not working well—people are getting lots of "Internal Server Errors" when they try to post comments, and I'm seeing the same problem when I try to post articles. Despite the error message, though, the comments are usually going through. IGNORE THE ERRORS.
Here's the usual run of affairs. You type in a witty and insightful comment as you always do, and then click "Post". Usually, this goes smoothly, the comment gets sent, the server tucks it into its database, and the page is refreshed to show your sweet, beautiful words in the thread.
Lately, something is balking after the server stores your comment, and it's showing you an error page instead of your sweet, beautiful words. Now here's the problem, and this is important. Do not ever do this: click the "back" button in your browser, and then click on "Post" again to resend it. Your comment is already there, and you're just piling up more copies of the same comment.
Here's what you should do if you get the error. Click the "back" button. Select your comment from the text box, and select "copy" from the edit menu. Then click on the "refresh" or "reload" button in your browser, and see if your comment was actually stored. If it was, you're done. If it wasn't, then you should paste your words back into the text box, and then click on "Post" again.
Remember, the error messages may be lying to you. Don't trust them. Verify.
If you've made umpty-eleven submissions of the same comment, don't ask me to delete them for you, either. When I go in to the control panel to tinker, I get those error messages on just about every operation. It's driving me a bit nuts, so please don't add to it.
Comments
Actually, if this helps your tech department, I'm not even getting the error message. I hit "Post," IE tells me it's opening the page, and then nothing happens. It just stays on the same page. If I wait a bit and hit refresh, my comment does show up.
Posted by: Fatboy | May 11, 2007 2:13 PM
I received one of those 500-Errors yesterday, and rather than hit back to post again, I just opened a new tab to see my comment had gone through.
Also, sometimes I click "post," and nothing happens, and since I am an exceptionally weak person, I find myself wondering whether or not I didn't use enough of my precious strength to click the mouse button.
Oh well... Time for me to go do my daily push-up and feel the burn!
Posted by: Dan | May 11, 2007 2:16 PM
I was very embarrassed to have been taken in by this error message the other day. Because it ends up looking like "Ooh, I think my comment is so clever that I must post it eleven times!!!" Once I realized what had happened, it hit me that I should have verified that my comment wasn't posted before resubmitting it. Over and over. (Even though when blogger gives you an error message, it means it...)
To my credit I merely said you may feel free to delete the repeats, I didn't actively insist that you do it. ;^)
The wacky part is that I was sober the whole time...
Posted by: C. L. Hanson | May 11, 2007 2:50 PM
If you get the error, STOP. Open another tab or annother broswer window (depending on what browser you are using), navigate back to Pharyngula and the thread you were posting to, and check the bottom of the comment thread to see if your post is there. If it is, you can close the tab/window with the error.
Posted by: speedwell | May 11, 2007 2:54 PM
You would have to be a real ass to submit your post multiple times.
Posted by: Richard | May 11, 2007 3:43 PM
You would have to be a real ass to submit your post multiple times.
Posted by: Richard | May 11, 2007 3:45 PM
When I have posted it usually lags and stalls... but it ALWAYS posts the comment.
Posted by: Steve_C (Secular Elitist) FCD | May 11, 2007 3:49 PM
Nothing can go wrong ... go wrong ... go wrong ... go wrong ...
Posted by: John Pieret | May 11, 2007 3:51 PM
It always seemed to me like it would be trivial for the software to detect when somebody is attempting to post a comment that is 100% identical to a previous comment. Like most programmers, I'm more than happy to backseat-drive here and say that they should keep a table of IP addresses and hashes of the posts that they've posted and simply discard multiple posts with the same hash and IP address. The IP address isn't even necessary unless you want different people to be able to post the same thing ("me too!"?). Easy solution to a common problem.
Then again, I try to avoid writing software that interacts with people directly. It's a surefire way to assault your otherwise beautiful program with screwy input and unexpected behavior.
Posted by: Troublesome Frog | May 11, 2007 4:34 PM
Yes. People should be eliminated. ;-)
Posted by: Kseniya | May 11, 2007 4:37 PM
Kseniya,
My thoughts exactly. Writing drivers and programming microcontrollers and such is a blast. You can write fairly complicated software that does really cool stuff and can handle just about anything thrown at it. In many cases, you can list out all of the possible input conditions and handle them. On the other hand, soon as you try to write a basic text editor with a GUI, a user will show up and find a way to completely hose your system that you never could have dreamed up. I've come to the conclusion that no interactive application is safe from the average human. That's why I avoid them at all costs.
Posted by: Troublesome Frog | May 11, 2007 5:07 PM
TF and Kseniya--yes; humans are poorly optimized. :)
Posted by: RavenT | May 11, 2007 5:09 PM
WORD! :-)
Posted by: mojojojo | May 11, 2007 5:17 PM
do you have access to the web server logs? they will give better descriptions of the error so it can be corrected.
Posted by: snex | May 11, 2007 5:19 PM
Maybe so - but can a wireless networking device remotely induce the canine defacation response?
Posted by: Kseniya | May 11, 2007 5:19 PM
"Umpty-eleven?"
I think you made Bad Astronomy Dude cry. ^_^
Posted by: tikistitch | May 11, 2007 5:26 PM
Are you sure the comment system wasn't designed by Discovery Institute programmers? What with the misleading messages and bizarre behavior and all.
Posted by: Rob Adams | May 11, 2007 6:43 PM
No, I believe that there is at least one thing humans (or *a* human) can do that computers never will :).
Posted by: RavenT | May 11, 2007 7:09 PM
Thanks for clarifying the situation for us.
Posted by: Coin | May 11, 2007 7:56 PM
What is it with people and the belief that humans are somehow magically capable of performing feats that other entities cannot?
Posted by: Caledonian | May 11, 2007 7:57 PM
Oops, I forgot to add the "joke" tags for the low-level parsers.
Posted by: RavenT | May 11, 2007 8:50 PM
It always seemed to me like it would be trivial for the software to detect when somebody is attempting to post a comment that is 100% identical to a previous comment.
Th Haloscan commenting software does that. When it decides (frequently) to have a nap, it rejects completely novel posts as duplicates. Then, once you make a minor change and resubmit it, it tells you that you've lost your cookies. When you try again, because you can see that your cookies are fine in the display, it tells you you can't post again so soon, and please wait another -29 seconds. After you wait a couple of seconds and try again it tells you that the service is unavailable. At which point you close the window down and re-open it, and start all over.
What is it with people and the belief that humans are somehow magically capable of performing feats that other entities cannot?
I really don't want my computer asking to borrow my KY before it does an image search of mainframes, capice?
Posted by: Graculus | May 11, 2007 9:01 PM
#'s 11 and 12,
If you try to make something idiot proof you can be sure someone will come along and make a better idiot. Which, BTW, is why no piece of software will ever pass the Turing test.
Posted by: Fernando Magyar | May 11, 2007 10:43 PM
No, the Turing Test doesn't require that the software make intelligent conversation, just that it's enough to convince the conversee that they're talking to a human being.
After a while, the Turing Test's standards have to be raised to the point where a human being can't pass it - and that's the point of victory.
Posted by: Caledonian | May 11, 2007 10:57 PM