[OS X TeX] Skim SKAutoReloadFileUpdate

[Adam M. Goldstein]

On Dec 19, 2008, at 12:54 PM, Enrico Franconi wrote:

> On 19 Dec 2008, at 18:42, Adam R. Maxwell wrote:
>
>>> So, it seems to me that the auto feature is the best compromise to  
>>> have a smooth workflow.
>>
>> If you want to use a brute force reloading and hope that corrupted  
>> data structures are dealt with correctly everywhere in Apple's  
>> frameworks (they're not), go for it.  Personally, I think it's a  
>> foolish approach when the problem is easily avoided.
>
> I agree with you in general, but from the discussion in this thread  
> I had evidence of the contrary: this is not *easy* in the case of  
> latex-generated files, since the workflow may fail for many  
> different reasons.
> Now, I also have learnt from the recent posting by Joseph that in  
> emacs+auctex (which is my scenario) there is already a way to detect  
> successful runs of the latex-ing process. I am willing to explore  
> this route, so that, as you suggest, I can at the end launch a  
> reload of the file.
> Does anybody know more about this possibility?
> cheers
> --e.

I don't know about it but as an emacs+auctex user I would be  
interested in testing it, etc., if someone else were to work it out.

In the meantime, I gather the following from this discussion:

(1) The method used by Skim to reload files causes a race condition;

(2) This condition can result in attempts to load incomplete or  
garbaged PDF files;

(3) Using TeXShop as a viewer won't help, because unless the typeset  
command comes from within TeXShop, it won't be able to tell when the  
typesetting process has ended, and this poses the same risks as using  
Skim.

(4) The risk of this is at worst, crashing Skim, or producing a  
partially-formed PDF, and this won't happen for garden-variety  
typesetting jobs (short articles or essays). The use of BibTeX and  
other additional runs of typesetting processes (indexing, I don't know  
what else) increases the risk of something going wrong because there  
is more regeneration of the file, making it harder for Skim to keep up.

(5) There is some way that Adam M. suggests might improve matters, but  
that some others think will be harder to implement.

(6) In principle, the auctex-emacs case can be solved, because auctex  
provides information about the completion of typesetting runs that  
could be used to trigger reloading only if the PDF is done being  
created, and is error-free.

At this point I am still using Skim as a viewer, since no problems  
have come up for me in practice.

-Adam G

------------------
Adam M. Goldstein PhD, MSLIS
--
agoldstein at iona.edu
a.m.goldstein at mac.com
http://www.iona.edu/faculty/agoldstein
--
(914) 637-2717
--
Dept of Philosophy
Iona College
715 North Avenue
New Rochelle NY 10801