On Mon, 28 Jan 2002 19:50:19 -0800, Greg Stein wrote: >On Mon, Jan 28, 2002 at 08:52:03PM -0600, William A. Rowe, Jr. wrote: >> From: "Brian Havard" <[EMAIL PROTECTED]> >> Sent: Monday, January 28, 2002 9:33 PM >> > On 28 Jan 2002 21:58:16 -0000, [EMAIL PROTECTED] wrote: >>... >> > > Add a couple new command types to process creation: >> > > >> > > APR_PROGRAM_ENV: start the program using the caller's environment >> > > APR_PROGRAM_PATH: search PATH for the program, use caller's env >>... >> > Wouldn't it be simpler to just have env==NULL imply using the parent's >> > environment? That is in fact already the case in the OS/2 implementation >> > (and Win32 by the looks of it). >> >> + 1 ... this patch [appeared] overly complex in the first place. >> With an env value passes that array of envvars. Without simply passes >> the current environment. > >It wouldn't really be simpler. If somebody accidentally passes NULL for the >env and it copies the environment down... oops. Security issue. In Apache, >we've taken some pains to limit the environment passed down to children. > >But it would also open up a combination that isn't possible -- >APR_PROGRAM_PATH with env==[...]. In words that is, "search for the program >on PATH, and here is the environment to use." That combination isn't >available on Unix, at least.
Well, other platforms CAN provide it and I'd guess you could do it on unix too if you really tried (do the path search yourself perhaps?). The non-unix platforms sometimes have to jump through hoops to implement behaviour that is standard on unix (usually because the API was designed by a unixhead :) so why shouldn't the reverse be true sometimes? -- ______________________________________________________________________________ | Brian Havard | "He is not the messiah! | | [EMAIL PROTECTED] | He's a very naughty boy!" - Life of Brian | ------------------------------------------------------------------------------
