I've got some example code on what I'm doing, however, as I expected, the error is gone! All involved objects were taken from the project, but not related code has been stripped-off. (Which means that some of that "at code level not related code" is in fact "related at compiler level". I still suspect references. Anyway, the code is still a bit rough, I'm going to do some further testing. The code is useless for debugging, but might give you an idea on what I'm doing, if you like i could make it available for download. On 11/23/05, Jay Blanchard <jay.blanchard@xxxxxxxxxxx> wrote: > > [snip] > I understand your point, but I can't offer you the project as it's company > property. Again, I will try to make a reproduce/example script. > Meanwhile, try to believe my assumption that the PHP compiler is broken > because: > [/snip] > > Have you thought about sending your description of the problem to the DEV > list? Again, they are going to ask for code snippets, error messages, > items > from the log. Code snippets or explicit examples of the code, while in the > pursuit of solutions, would probably not violate any company > ownership/rights issue. > > [snip] > Still, to rule out any mistakes of mine, I tried the code at a > ZendPlatform enabled server and the bug seems to disappear. Not really, > though, because a value seem to leak to someplace else and even worse: > while > ZendPlatform seems te work perfectly (I can debug everything, even this > specific piece of code) httpd segfaults when debugging this code with the > "evil" parameter > [/snip] > > Again, without seeing the specific piece of code, as a starting point, it > would be hard for us to speculate. > > [snip] > - Back to the error message: even the error message is corrupted, the > compiler doesn't seem to know what it's doing (what would explain de httpd > segfault with ZendPlatform: PHP does something it's not meant to do) > [/snip] > > In what way is the error message corrupted? > > At this point we can only speak in hypotheticals. Assuming your code is > correct the compiler is broken somehow. Without knowing what the code is > we > cannot begin to know how the compiler is broken. > >
