On October 22, 2002 11:04 pm, Yasuo Ohgaki wrote: > Ilia A. wrote: > > On October 22, 2002 10:30 pm, Yasuo Ohgaki wrote: > >>Ilia A. wrote: > >>>>What's the point of _not_ specifying certain ini for > >>>>run-tests.php? In other words, what is advantages? > >>> > >>>Because those ini settings may be the ones causing the problems. Unusual > >>>ini settings could lead to discovery of various problems, which may not > >>>show up under 'normal' configuration. > >>>We are trying to determine if the compiled PHP is working properly with > >>>the user's settings and system configuration, not just with predefined > >>>settings. > >> > >>No augment for that. > >> > >>ini used by run-tests.php has nothing to do with the ini > >>used by phpt. It seems there is confusion about this. > > > > That is completely true. However, while the test scripts are mostly very > > basic 'theoretical' tests of various functions, run-tests.php is a fairly > > complex script with a practical application. IMHO, this script itself is > > part of the test. Should it fail for whatever reason, that would indicate > > some 'break' due to ini settings, allowing the user as well as the > > developers (should a user submit a bug report) to determine if there is a > > problem and handle it accordingly. > > I agree. We need real complex tests. > However, I have different opinion. We do need complex tests, but > we don't want to people tells us "Hey it does not work" merely > slight ini setting difference _for_ run-tests.php.
Complex tests would be nice, but I doubt many admins would be willing to wait 10 or more minutes while a complex test suit runs. And given the fact the PHP test suit had matured greatly, especially in 4.3.0 many admins may use it to determine if the compiled source is ready for 'prime time'. If a particular ini setting is casuing a problem, such as an error reporting level set to high or to low we should be aware of this issue and address accordingly. This is by either making the code compatible with various ini settings or setting them via the -d parameter before execution. This is not a hypothetical issue, just today I was able to track down a slight bug in our EOL detection mechanism because I had auto_detect_line_endings turned on. Had I been using the ini-dist, which has this option disabled I, would not have seen this problem. > The fact tells us that even we don't understand which setting > is affecting what. Therefore, I think we are better to specify ini. > It's a lot easier for us. Well, it is generally fairly easy to track a ini related problem to 2-3 suspicious settings. And being aware of the problem is the 1st step at solving it, with the ini-dist the problem may simply slip though and later appear in user's production code making it infanetely more difficult to trace & debug. > > e.g. Do we want to make run-tests.php run well under magic_quote_runtime > , magic_quote, etc enabled environments? As I've said before, certain options such as magic_quote_runtime can be turned off, or we could simply modify our script to be aware of the effects this option may have and handle itself appropriately in the event this option is on. Ilia -- PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
