[Orgmode] Re: doc-view-mode

[Óscar Fuentes]

Hello Tassilo.

Tassilo Horn <address@hidden> writes:

>> Converting the pdf|ps|dvi file to images is exactly what your
>> favourite viewer does.  The problem with doc-view-mode is that it
>> converts *all* pages on the document to image *files* which are left
>> on some place of the hard disk.
>
> I think that's the only practical solution, else you would have to wait
> about a scond when switching to the next page.

I have no problem waiting a second or two, although some experimentation
shows that it is pretty fast opening a small pdf doc and even faster
refreshing the image after M-x doc-view-enlarge. This is on Linux with
a 2.4 GHz CPU.

[snip]

> Why are many files a problem for you?  With that approach, opening a
> document for the second time works instantly.  And by default those
> images are stored somewhere in /tmp:

As said above, instantaneous response is not required for me. The
problem with converting all the document to image files is that I often
open large pdf's (several thousand pages) and small/medium dynamically
generated pdf's. This would create tens of thousands of image files on a
busy day (no exaggerating) which would require several gigabytes.

Perhaps the most important problem with converting the full doc to image
files is that it is a cpu and filesystem intensive process that can take
a long time (think a fully illustrated 1000 page pdf). It steals cpu
cycles on a busy machine (I often read while a long build or test suite
is running) and drains battery on laptops/netbooks.

[snip]

IMHO, a user-configurable switch for "render this page and delete it
before rendering the next" would be okay. More advanced options like
keeping just the last N recently viewed pages of M documents (plus the
succesive page of the current one) would be nice too, but if
doc-view-mode supported the simple one-page option, it would be fine for
me.

P.D.:

After some experimentation with doc-view it seems to me that the file
image cache system is flawed: open a large pdf file -> doc-view starts
conversion -> enlarge -> doc-view cancels previous conversion, throws
away the files, and starts a new one -> shring -> cancel, throw and
restart again, etc -> close the pdf view -> open the same pdf view -> if
the cache I guess you thought that creating separate caches
for every new zoom level would be too much caching :-)

It seems there is a bug: open a large pdf -> before the conversion ends,
kill the buffer -> open the pdf again -> the conversion is not resumed,
only those pages converted on the previous session are accesible.

-- 
Óscar