On 10/2/2013 2:15 PM, Norbert Marwan wrote: > dpb <firstname.lastname@example.org> wrote in message <email@example.com>... >> >> One would think it should be a bug indeed but unfortunately it isn't. :( >> > why not? > in my opinion this is a bug when it is not working as it should do. i do > not see any reason why the whitespace should not be counted when using > the format strings. > ...
Unfortunately it's not a bug because it _is_ working as specified/should do. It's just that the definition of what the width field means on input w/ blanks isn't consistent with what it seems intuitively it should be (or what Fortran FORMAT means).
You'd have to ask the authors of C the "why". It's what they chose to do on input scanning and since Matlab i/o is built around the C library functions, it (Matlab) behaves as do they (C i/o).
That it (as well as much else about C formatted i/o, but really more on input than output) was a sorry choice I wholeheartedly agree but it's done and shall never change now, unfortunately.
TMW _should_ however, build in another set of i/o functions that do follow Fortran-like behavior. Ideally these would also follow Fortran FORMAT syntax.
Again, file the enhancement request...unfortunately as previously noted I harped on this years ago and it never received a lick of attention. My license support lapsed and at that point TMW wouldn't even accept offered bug reports so I gave up the crusade to simply harranguing here in cs-sm when the subject arises (as it doesn't quite regularly).