On Mon, Apr 15, 2002 at 11:34:43AM -0400, Pierre A. Humblet wrote: >Christopher Faylor wrote: >>It looks like the patch will do the job but I would like to be >>convinced that there is no other way around this problem. If I'm >>reading this correctly, this change requires modifying any code which >>uses cygwin. That's something we should try to avoid at all costs. > >I second that 100%. My proposal *allows* porters to avoid CLOSE_WAIT >by making some changes in daemons, but none in the children processes. >No change is required anywhere. The additional Cygwin functionality >would be invisible to applications that don't explicitly use it.
How can you second that 100% and then talk about how people have to change ther code to accomodate cygwin? The fact that it is server code rather than client code really doesn't matter. Again, I don't like adding cygwin-specific features that mean mucking with source code that works perfectly on unix. The goal of cygwin is to avoid ifdef's and just allow configure/make/make install. It may well be that your solution is the best that we can hope for but I don't want to see it included in the source until I'm convinced of that fact. cgf
