Bug or feature in StandardFileStream

>
> The following behaviour seems odd to me:
>
> file := StandardFileStream readOnlyFileNamed: 'test'.
> file skip: self size.
> file atEnd.  (-> true, OK)
>
> file skip: 1.
> file atEnd (-> false ???)
>
> My VM is rather old (> 1 year), so maybe someone should test this with
> a more recent one.

> On Sun, Feb 12, 2006 at 12:02:21PM +0100, Martin Kuball wrote:
> > The following behaviour seems odd to me:
> >
> > file := StandardFileStream readOnlyFileNamed: 'test'.
> > file skip: self size.
> > file atEnd.  (-> true, OK)
> >
> > file skip: 1.
> > file atEnd (-> false ???)
> >
> > My VM is rather old (> 1 year), so maybe someone should test this with
> > a more recent one.
>
> It's a feature. It just means that you can set the file position beyond
> end of file, at which point it is no longer "atEnd". You can use
> #setToEnd to position the file pointer to the actual end of file.
>
> If it did not work this way, you would have no way to (for example)
> write a single byte at position 1000 without first having written
> something to the preceding 999 positions. You probably don't want
> to do this very often, but at least it's possible.
>
> It works the same on Windows and Unix, so I expect that the behavior
> is consistent across all platforms.
>
> Dave

> Le Dimanche 12 Février 2006 15:19, David T. Lewis a écrit :
> >
> > It's a feature. It just means that you can set the file position beyond
> > end of file, at which point it is no longer "atEnd". You can use
> > #setToEnd to position the file pointer to the actual end of file.
>
> Maybe it's a usefull feature for files opened for writing, I don't know.
>
> But Martin had its file opened readOnly, in which case such a feature makes no
> sense.
>
> What would file next will answer? an infinite collection of zeros like a null
> stream ?

> On Sun, Feb 12, 2006 at 03:38:34PM +0100, nicolas cellier wrote:
> > Le Dimanche 12 Février 2006 15:19, David T. Lewis a écrit :
> > > It's a feature. It just means that you can set the file
> > > position beyond end of file, at which point it is no longer
> > > "atEnd". You can use #setToEnd to position the file pointer to
> > > the actual end of file.
> >
> > Maybe it's a usefull feature for files opened for writing, I
> > don't know.
> >
> > But Martin had its file opened readOnly, in which case such a
> > feature makes no sense.
> >
> > What would file next will answer? an infinite collection of zeros
> > like a null stream ?
>
> It answers nil.

> > Everybody would prefer getting an Exception rather than this
> > strange behavior i guess.
>
> Agreed.
>
> Note the squeak file system code was written many years before the
> introduction of exceptions, and there is wide agreement that it is
> due for some rework. There is a team forming up to look into this:
>
>  <http://www.squeak.org/Community/Teams/>
>
> <http://lists.squeakfoundation.org/pipermail/squeak-dev/2006-Januar
>y/099050.html>
>
> Dave