Shapira, Yoav wrote:Well, this sounds quite alarming to me. Looks like there is no guarantee that the contents of the build really contains the intended codeline. Could some practices be established that would prevent such situation from happening?
Hi, I have a strange issue with 5.0.23 that's making me question my stable vote from earlier today. It has to do with a 3rd party servlet which checks for the presence of a cookie, if not found prompts for some user information, and then redirects (response.sendRedirect) to another page. The exact same code base and test case works in 5.0.19 but not in 5.0.23. Was something relevant changed?
Looking at the changelog, I'd like clarification on: "Optimization: delay parsing of the cookies (remm)"
I'm running Sun JDK 1.4.2 on Solaris 8 (all patches installed). As I mentioned it works with tomcat 5.0.19. I'll proceed to try the intermediate builds 5.0.20, 5.0.21, and 5.0.22 to further narrow the possibilities. Thanks,
I did find what the problem was: I did some testing, so I checked out an earlier revision of the file (CoyoteRequest, of course). Unfortunately, I'm using a new CVS tool with which I have less experience (tortoise) - not that I wanted to, but WinCVS decided to simply stop working one day - which doesn't display tags in a visible way, and the result is that an older revision of the file is in the build :(
For example:
- use command line CVS to check out sources for production builds
- for production builds, always do a clean check out (use an empty directory)
- write an Ant script that checks out sources for production build
Would this help?
Thanks Petr
I'll have to do a new 5.0.24 build today, then.
Rémy
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
