Re: Re: Checklist
| From: | Yasuo Ohgaki | Date: | Tue, 05 Mar 2002 07:18:14 +0000 |
| Subject: | Re: Re: Checklist | ||
| References: | 1 2 | Groups: | php.qa |
| Request: | Send a blank email to php-qa+get-4459@lists.php.net to get a copy of this message | ||
derick@php.net wrote:
On Mon, 4 Mar 2002, Yasuo Ohgaki wrote:Since I create bug reprot by my self before fixing, all problems that I know or fixed are in bug db. So basically we should test bug reports closed as a regression tests. If my patch is applied, we can test - crash with invalid sesessin id, session name, save path, - crash under very high load (mm) - crash under user save handler - memory leak fix with invalid number of paramters - many fixes I don't even remeber now :) Outstanding issues that is critical - crash with $_SESSION/$HTTP_SESSION_VARS (there is still) - invalid reference for $_SESSION/$HTTP_SESSION_VARS makes $_SESSION or $HTTP_SESION_VARS uselessDerick Rethans wrote:Some of them, but would you be so kind to write a few lines with sugested testing for QA here? So that others know what to check exactly, and see whether it crashes or not? This makes life for QA ppl a little bit easier. (It's not interesting now how it failed, just what could fail).Hola, here is a list that falls into the 'special' area of testing: - sessionsYou know issues.
Basically all. I have touched all function to change function names. :) There are lines that are changed to support async functions. Older functions are very tested well, though. New functions should be tested more.Ok, I know, that's why I listed it here :)- macosx build - ODBC - overload - HTTP file uploads - output buffering (especially in combination with gz_handler and mb_string goodies)mb_output_handler is broken now. It should be easy to fix, but I haven't take a look at it yet.FYI: flush feature for nested output buffer must be implemented. My patch does not implement it, but it just disabled broken flush implementation. Work around is disabling output buffering ;) This is one of the reason why I force "make test" to use php.ini-dist.I still do not think that is a wise idea, and when exactly does the flush not work?- pgsql pgsql extension has a _lot_ of new code and many lines are modified. I tested, but others should test the code also.Could you write down what exactly should be tested here?
Referer? This has been fixed, thoguh. I build ZTS with Apache2 SAPI. I don't have problem, but not many people are testing with multi thread server SAPI. Therefore, we are better to focus a little more. We are confident that there are more bugs with ZTS build than normal build, right?- ZTS build.What's up with this? I usually do ZTS builds and normal builds, and had no single problem with the ZTS builds.
There are many changes in between 4.1 and 4.2. Output control code are different (Zeev and Me) PostgreSQL has a lot of new functions (Me) mbstring has now multi-byte regex functions and function overload feature (Rui) etc, etc. Basically, you would like to know you should check NEWS file and check related commited for more details. (and source) Or are we going to ask developers what has been changed? I guess not :) -- Yasuo Ohgaki- building with CLI less important: - NULL as 2nd parameter to pg_fetch_* - DIO - preg_grep (nice function name BTW :) - pow() - domxml - tokenizer extension - iconv