Bug Fixes - that yellow bastard

recent entries:
friends | friends2:
archive:
my friendfeed:
about me:



tags
about me advice art atlanta being yellow books cars climbing comics dad fic food games gaming geekism go gators hell is... holy crap home insomnia language meme money movies music odd poetry poker politics quotes singletude spam tech this sucks travel tv why i rule why i suck work

more bastard
bronze vip archives
notes of a code poet
friendster
furious ming
del.icio.us
dude check this out
pandora
that bastard multiples

that bastard suggests
nandot
cap'n ken
coffee achiever
el guapo
bad news hughes

the stack
secret history:

the queue
battlestar galactica::

recent posts
+ khubli
+ skander

August 29th, 2003


Previous Entry Share Next Entry
2003.0829.1139::Bug Fixes

Martha, my co-worker/supervisor, and I have been banging our heads against the wall for over two weeks about a bug in the system. Essentially, we've got a homegrown Python-based web server that serves content via localhost. Well, all of our Win32 installations were crashing if the user were to enter a quick Pythonserver-Form Button-Pythonserver cycle. Somehow, it'd choke on requests, and end up crashing in a way that it'd only work on non-form-button requests.

Martha finally found the problem today--because Win98 Internet Explorer somehow chokes on the way we're delivering the Content-type (it actually flashes to a page with just the Content-type displayed, and then forwards to the actual returned page). Windows 2000 and XP both handle it fine--it's just Windows 98. But since the original code was written for 98, we didn't foresee that the original bug "fix" would affect the new versions. Sigh.

Anyways, along the way, we discover:
  1. When a C-invoked Perl process (via the Win32 CreateProcess()) makes the system call for `ver`, Windows 98 returns nothing. Win2k and XP, however, act normally.
  2. Internet Explorer is perfectly fine handling html content without an actual Content-type field.
  3. Internet Explorer is largely fine handling html content without valid HTTP headers (said headers can be either " "--yes, that's a blank space--or a Content-type with just a single newline--no, there are no other headers to add a second newline).


But, thankfully, it's all fixed now. I wonder how many dumb Microsoft workarounds have seriously screwed people over when using later versions of their products?

Raise your hand if you want Kill Bill to be Quentin Tarantino's documentary about Microsoft...

And in a geeking out of another kind, I'm off to Dragoncon at noonish--I'll be the tall Chinese guy with the bright orange Blue Man Group shirt, hanging around the Buffy track. Say hey if you care to.
Current Mood: [mood icon] relieved

2 comments | Leave a comment )

Comments:


khubli::2003.08.29.09:17 am
[User Picture] hey geeky chinese guy in orange blue man group shirt... funny i'm wearing a bright orange REI superwicking t-shirt today... must be an Asian thing.. I don't quite understand
skander::2003.08.29.11:08 am
[User Picture]Please don't get me started on IE behavior. We ran into this crap with our website.

I love this part:

"However, because most developers are unaware of this request style, they treat each GET the same and return the entire document."

Maybe the developers are unaware of that request style because YOU DECIDED TO PULL IT OUT OF YOUR ASS AND STARTED USING IT.
Go to Top:
LiveJournal.com