ejd@iris.brown.edu (Ed Devinney) (12/12/89)
Given that PenPixPat() and BackPixPat() actually place the PixPatHandle passed to them into the pnPixPat or bkPixPat fields of the CGrafPort, it seems reasonable to clone the PPhandle before passing it to one of these routines, as CQD seems to dispose of the handle when it changes the pn/bkPixPat. Fair enough - better that than have the program forget about them and eat up heap space. However, we're doing A/UX development here, and the memory handling is less than stunning; repeated allocations of the sort needed to clone PPhandles seem to fragment the heap (which in A/UX does not seem to compact/move). Any ideas out there for sound ways to reduce or eliminate the large quantity of allocations that we seem to have to make? Thanks for any help y'all might be able to provide. ed Ed Devinney ... ejd@iris.brown.edu ... Providence, RI "I don't know about your brain, but mine is really...bossy" - Laurie Anderson
tecot@Apple.COM (Ed Tecot) (12/17/89)
In article <22616@brunix.UUCP> ejd@iris.brown.edu (Ed Devinney) writes: > However, we're doing A/UX development here, and the memory handling is >less than stunning; repeated allocations of the sort needed to clone >PPhandles seem to fragment the heap (which in A/UX does not seem to >compact/move). > Any ideas out there for sound ways to reduce or eliminate the large >quantity of allocations that we seem to have to make? Thanks for any help >y'all might be able to provide. I'd suggest that you not worry about it. A fragmented heap isn't really a bad thing under A/UX, since it has a virtual memory system beneath it. _emt