RFR 9 8055330: (process spec) ProcessBuilder.start and Runtime.exec should throw UnsupportedOperationException ...
martinrb at google.com
Fri Jan 30 23:41:36 UTC 2015
It often happens that some ancient thought was in my head 10 years ago, and
it takes a day or two to get paged back in.
Now I think that 10 years ago I definitely considered the possibility of an
OS that had no subprocess support at all, and I was comfortable with simply
throwing IOException, much as if whatever command you were looiking for
could not be found. Maybe there are no files in the file system... Right
now I'm thinking a sandboxed OS should throw IOException, in part because
UOE is not documented, while IOException is. Nothing is stopping you from
throwing a new subclass of IOException, e.g.
YouAreTrappedInASandboxException extends IOException ...
On Fri, Jan 30, 2015 at 1:48 PM, Roger Riggs <Roger.Riggs at oracle.com> wrote:
> That does read better.
> As in this webrev:
> On 1/30/2015 4:30 PM, Alan Bateman wrote:
>> On 30/01/2015 15:58, Roger Riggs wrote:
>>> Please review this clarification to the optional behavior of
>>> java.lang.Runtime and java.lang.ProcessBuilder
>>> on platforms that don't support process creation.
>> One suggest is to move this to after the "Starting an operating system
>> .." paragraph. The reason is that it more logically following the list of
>> highly platform specific issues. Otherwise looks okay to me.
More information about the core-libs-dev