From: Gil B. <gba...@al...> - 2009-03-30 17:53:17
|
Regardless of whether or not the prologue code in ::requires files was or was not a hack, it seems to me that its use may be fairly common. The ooDialog code, Mike's programs and BSF4Rexx all seem to have run into this already and who knows how many other users will experience a problem when the code becomes more generally available. It does not seem reasonable to me to require everyone to rewrite their ::requires files in order to use the new version of ooRexx. Or did I misunderstand this thread? Rick McGuire wrote: > Yes, that's pretty much the case. A lot of those uses were hacks > designed to get around the deficiencies of the old loading model. > > Rick > > On Wed, Mar 25, 2009 at 2:06 PM, Mark Miesfeld <mie...@gm...> wrote: > >>On Wed, Mar 25, 2009 at 10:05 AM, Rick McGuire <obj...@gm...> wrote: >> >> >>>The users *might* have gone away...there could be references to the >>>package on instances running on other threads. The timing of when the >>>weak references can be cleared are subject to the dynamics of the >>>garbage collector. One factor that might result in it not getting >>>cleared immediately upon termination is uninit methods on class >>>instances created from the same package. It's really just a matter of >>>timing on when that gets cleared. So if the package manager still has >>>an instance in the cache, it gets used. The cache entry can >>>potentially get cleared out to prevent a persistent memory leak, but >>>the weak reference will not get cleared at the exact moment the object >>>goes out of scope. It takes at least one GC cycle (potentially more >>>if uninit methods are involved). >> >>That all makes sense to me Rick. And, I am far, far, more in favor of >>the current behavior of the package only getting loaded once than the >>old behavior of the package getting loaded over and over. >> >>So, for people using the C / C++ APIs to start the interpreter more >>than once in a single process, the advice would be to not use any >>package files that rely on the code before the directives being >>executed? >> >>-- >>Mark Miesfeld >> >>------------------------------------------------------------------------------ >>_______________________________________________ >>Oorexx-devel mailing list >>Oor...@li... >>https://lists.sourceforge.net/lists/listinfo/oorexx-devel >> > > > ------------------------------------------------------------------------------ > _______________________________________________ > Oorexx-devel mailing list > Oor...@li... > https://lists.sourceforge.net/lists/listinfo/oorexx-devel > -- Gil Barmwater |