[Okular-devel] Okular auto center papers, any way to prevent this?

[Jochen Trumpf]

Hi,

I think I agree with Luigi.

Luigi Toscano wrote:
> I think that you are talking about two different things.

The first one is the behaviour when going to a different page (e.g. by going 
to the next page.

> Mark Altern wrote:
> > Sure, there is something like: < 5 of 34 > at the bottom of the page,
> > and when you click >, it turn to next page, but every time you click it
> > to turn to next page, it will auto-center the page.
> 
> This is what all the viewers I tested do: kdvi, evince, acroread, xdvi
> (and okular, of course). When you change the page, the view changes to
> the upper-left corner of the chosen page. This is a consistent
> behaviour, I don't think it's worth to change that.

Except that acroread goes to the upper-left corner, while okular centres the 
page horizontally. If, for example, your zoom level is such that the page is 
much wider than your okular window and you have been looking at a non-central 
portion of the page previously, upon switching the page the horizontal  
position will go to centre of page. In acroread the horizontal position will 
go to the left-most portion of the page (or the right-most in RTL texts).

So, both these viewers change the horizontal viewport, but differently. Not 
sure which one is better.

I can see a use case for not switching the horizontal viewport (e.g. comparing 
margin decorations across pages) as well as for switching it to the left (e.g. 
reading two-column layouts). I am not so sure what a good use case for 
centring is, but maybe to get an "overview" of the new page?

> > In most cases, the dvi file I am viewing is not-symmetric and it is
> > zoomed in as large as possible to fit my screen, thus the pages are NOT
> > centered. Then when I use emacs to forward jump from emacs to okular, it
> > will jump to the proper line in proper page, BUT it will auto-center the
> > page that makes the page land in the middle, which is of couse will
> > messed up its original hozinontal position.
> 
> What is that "forward jump", exactly? Have you configured emacs to call
> okular and jump to the specified line? Can you provide us this
> configuration?
> I don't think that this is the same "issue" you described before,
> because as I told you, kdvi acts exactly like okular, in that case.

This seems unrelated. I believe Mark is talking about forward search in dvi 
documents. This still has a few bugs. In particular, last time I looked, the 
forward search routine was not using horizontal information to determine the 
closest source anchor. This would explain why the horizontal position is not 
properly matched in forward search.

There was a brief email exchange with Pino regarding this a long time ago, but 
I don't think it has been changed (read: I couldn't be bothered to produce a 
patch and nobody else could either).

http://www.mail-archive.com/okular-devel@kde.org/msg01470.html

Cheers,
Jochen