On Fri, Mar 12, 2010 at 04:49:37PM +0100, Thomas Wolff wrote: >On 12.03.2010 16:16, Eric Blake wrote: >>> ... >>> >> This is an area of active conversation; if you would like, you can test >> the latest snapshot and the experimental coreutils 8.4-1 to see if the >> behavior is more intuitive (that is, there are more situations where >> .exe is preserved across file moves or copies, and fewer places where >> .exe is appended on a whim if the source didn't have one). >> >> In general, cygwin does not care if the .exe is missing, but other >> programs (particularly cmd) do, so it is better if PE-COFF files are >> given the .exe extension. But implementing it is tricky - for example, >> in the case of 'cat a> b', there is no way to tell at the time when b >> is created whether it will be populated with PE-COFF contents (that is, >> no way to tell whether the source was a literal 'a' or 'a.exe'), so you >> will not get an .exe in that case. >> >I'm just pondering the bold idea (probably to be discarded) that it >*could* be detected by "magic number" checking, i.e. renaming a new file >on-the-fly after a few bytes of PE-COFF have been written to its >beginning... 8-)
...which is pretty much what we're already doing... cgf -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple