views:

417

answers:

3

Hi, I'm using cURL and php to query an exchange server that has Forms Based Authentication turned on. Due to FBA being on I need to collect a user's cookie authentication ids first. Yesterday everything was working: then something changed overnight and stopped working. I noticed that the cookie ids I was getting back were different (not the ids themselves because they change with every new call. Rather, the format and other things changed).

I use to see this:

# Netscape HTTP Cookie File
# http://www.netscape.com/newsref/std/cookie_spec.html
# This file was generated by libcurl! Edit at your own risk.

sdgh.utest.edu  FALSE / FALSE 0 sessionid aeddf34c-16fe-4126-8c96-da7a33d244b9
sdgh.utest.edu  FALSE / TRUE 0 cadata "0x0zHQHA3o8Vcu/DK/MWp69xkkZAJkIMts3HrOeHc4a71GN0CbqnDLiadJfKfSsseiyyxkyyUhRQ="

And now I see this (notice the url on the second line and the "#HttpOnly_" in front of the 2nd id):

# Netscape HTTP Cookie File
# http://curl.haxx.se/rfc/cookie_spec.html
# This file was generated by libcurl! Edit at your own risk.

sdgh.utest.edu  FALSE / FALSE 0 sessionid 11358a02-233c-4425-8a83-6b8b69b21bcd
#HttpOnly_sdgh.utest.edu    FALSE / TRUE 0 cadata "0FHKgxeq7st0DW57zl5HDgZ5gp7eRDqgfG6iAyxSf4TvyJI5aZI5wmPeiEHFloGo2bvH0fpnl52c="

Can anyone tell me what this sudden change indicates?
Did something change on my web server?
Could it be the exchange server that I'm querying?
Did someone update cURL on me behind my back overnight?
Also, I do not have automatic updates happening on my web server.

Does anyone think this even implies that cURL was updated? I didn't update it as far as I can tell. When/why does cURL or libcurl switch from http://www.netscape.com/newsref/std/cookie_spec.html
to this
http://curl.haxx.se/rfc/cookie_spec.html

Thanks for the help!

A: 

Do you have automatic updates enabled?

Maurice Perry
Not on my web server.
Chain
More clearly, No I do npt have automatic updates enabled on my web server. Theoretically something could have changed with the exchange server since I don;t control it.
Chain
A change to curl seems more likely to me: why would an exchange server send the URL: http://curl.haxx.se/rfc/cookie_spec.html ?
Maurice Perry
Thanks Maurice, that helps a little. I'm hoping someone can tell me when cURL might have switched from the netscape url to the haxx.se url then, if it's true that cURL outputs that url. If I can find that out then I can track my problem down.
Chain
A: 

Just to make things clearer: Something changed in the way that the cookie is being stored: If you saw your collected cookie data go from http://www.netscape.com/newsref/std/cookie_spec.html
to this
http://curl.haxx.se/rfc/cookie_spec.html
(and also, the "#HttpOnly_" part in front of the second cookie ID)...

Would you think that there's any chance that a change on the exchange server could have caused it... Or would you think that it was 100% likely that it was your own web server that changed?



I searched all around the internet and haven't found any more info. Does anybody suspect that cURL stopped using the "netscape" url at a certain version number?

I'm flat out confused as to how this switched on me. Any addition thoughts would help at this point.

Thanks.

Chain
+1  A: 

Ok,here is an explanation for the #HttpOnly_ prefix: http://blog.php-security.org/archives/40-httpOnly-Cookies-in-Firefox-2.0.html

My understanding of the issue is that curl or libcurl was updated and that as a result, the server now sees it as a user-agent capable of dealing with #HttpOnly_ cookies.

Maurice Perry