Re: Clairification - was [Re: Desktop Background Bites the Dust]
On Sat 27 May 2017 at 09:52:45 -0500, David Wright wrote:
> On Sat 27 May 2017 at 12:32:06 (+0100), Brian wrote:
> > On Fri 26 May 2017 at 17:57:40 -0500, Michael Milliman wrote:
> > > And actually, the --no-xinerama flag is not needed, at least on my
> > > system. I use feh --bg-fill ... and it works just fine. Of course,
> > > that is a very limited (and safe) usage of feh. I don't think it wise
> > > to use it for anything much more (IMHO).
> > Surprising after all these years that feh is now revealed as an unsafe
> > image viewer.
> Wow, I didn't realise I'd participated in a revelation.
It happens all the time on -user. :)
> I noticed the overwriting problem in September 2007, within days of
> buying a digital camera that set the orientation flag. (Previously,
> I'd been using a work cast-off which didn't set it.) I'd used xzgv
> for some years as my image viewer but never had much luck with its
> --exif-orient option, so I tried feh as an alternative because
> it appeared to honour it.
> Very soon after using feh, I noticed that the timestamps displayed
> by mc on my camera picture directories were getting screwed up.
> Of course, these picture files get backed up straight from the SD
> card before I even look at them, so I was able to restore the
> original pictures without any difficulty once I'd worked out
> what to blame.
My first reaction on encountering the overwriting a short time ago was
to be annoyed; perhaps a similar reaction to yours. After reading that
the transformation was lossless (no pixels harmed in the process) I came
to terms with it for a while. I regarded the original photos as having
the wrong orientation and didn't mind ending up with ones which were
"correct". Also, any file change does not materially affect printing.
Then I decided it is usually not a good idea to lose an original file,
so I unbound the editing keys and rebound them to backup the original
before conducting any rotations etc. The "Q" key was also rebound to
restore the file on quitting feh. This effort made me feel better but
I still wondered whether losing the original was technically critical
in all circumstances. I am persuadable.
If libjpeg-progs (libjpeg-turbo-progs) is purged the issue becomes a
> (However there are still probably a few downloaded images that are
> not in their original state, because I have no practical way of
> discovering which ones they are.)
> Apologies for not bringing it up then. I didn't resubscribe
> here for another 8 years and was only reminded about the problem
> on seeing this thread. FWIW I use xsetbg (xloadimage) for my
> desktops, but my laptops use the background colour as a
> battery-state and, in extremis, CPU temperature indicator. So,
> for example, if the power cord dislodges, the presently green
> background will switch to yellow.
The only backgrounds I contemplate having are ones with a solid color.
So its xsetroot for me on most occasions.
> > I use it to look at an image prior to printing it from feh
> > and see that as quick, reliable and convenient but not as unwise.
> Understood. I didn't know whether you were a feh user, or had just
> read the man page, when I made my mouse comment. In any case, such
> single-use instances as proof-checking and wallpaper-setting may
> well justify its use, but my comment about its misdescription
> and dangers was intended for readers of this list who might use
> it as a general-purpose viewer after installing it in the wake of
> this thread.
It doesn't take much effort to damn a piece of software and labelling
it as unsafe without a justification is one way to steer users away
from it. feh would probably now be "a thing of history" if it were not
for its present maintainer, Daniel Friesel. Most people would not read
the issues on github and could form an unjustified impression of feh
and its development from such unsubstantiated claims. Articulated
criticism is one thing, assassination by innuendo is another.
> BTW I checked the error message I mentioned earlier. Feh itself
> emits the incorrect error message for PNG files, but the correct
> message is emitted by the spawned jpegtran program for JPGs.
Incidentally, your mouse remarks (re Ctrl+2) are valid but (untested) I
expect the action could be rebound to a different button. Stretch has a
--auto-rotate option, which is very useful. It can easily be installed
on Jessie with a few extra packages to get it going.