From: ville.mattila@... Date: 2006-01-12T14:41:44+09:00 Subject: Re: 64-bit Solaris READ_DATA_PENDING Revisited >From: news >Date: 11.01.2006 22:11:34 >Subject: Re: 64-bit Solaris READ_DATA_PENDING Revisited > >ville.mattila@stonesoft.com writes: >> Hello, >> >> here is the updated patch. With ChangeLog entry. I've tested this with >> amd64 and all may tests >> passes. Yay! >> BTW, Would samekind of bug be reason for MAC OS X drb hang? In other word >> READ_PENDING gets >> defined uncorrectly? > >I was thinking something similar in [ruby-core:6983] although I was on >the wrong track then. I finally narrowed it down by looking at >syscall traces of sample/svr.rb. 32-bit Ruby was calling select() >with inifinite timeout, but 64-bit Ruby was calling select() with 0 >timeout[1]. So in the latter case it was falling through to accept() >immediately. > >Do you have a reference for the OS X hangs? Anything that causes >READ_DATA_PENDING = !feof() should cause the problem, it's not really >a 32/64-bit issue. I was thinking exactly the 6983 post 8). - Ville