[texhax] dvi vs pdf

[Pierre MacKay]

On 12/2/2010 10:40 AM, Michael Barr wrote:
> Here is my reaction to the various replies I have received.  First, as 
> long as TeX is TeX it will be required to produce dvi files.  As for 
> pdf, I have no doubt that when it is no longer supported, there will 
> be converters.  But the journal I help edit has several hundred papers 
> in pdf format (most also in dvi and ps).  Who will convert them.  It 
> would be a huge job and we run on zero budget.  To see what can 
> happen, what would you do with a Word file from, say, 1990?  Each 
> version of Word can handle files from the previous version or two, but 
> no further.  So unless that file was continually updated from version 
> to version, it is effectively dead.  I don't expect this to happen 
> with pdf, but it illustrates the problems.
>
> I have one further problem with pdf.  I am accustomed to compiling a 
> file, previewing it, making changes and then iterate.  I cannot do 
> that with the Adobe reader.  I cannot write a new pdf file while the 
> old one is loaded. So exit the file from the reader, then compile the 
> new version, load the new file, and then find your place again.  I 
> find this intolerable. Someone once suggested a different pdf reader 
> and I tried it and found it unsatisfactory in other ways (I no longer 
> recall why).  I'll stick to dvi, thank you.
>
Hear Hear!!!

I think I make about 50 passes through DVI for each article I prepare 
for the journals I set.  Many of them are at the last stage---page 
balancing---which often depends on adjustments as much as five pages 
earlier than the bad page break I am correcting.  
\stretchnextspread{<skip amount>} or \shrinknextspread{skip} or even 
\stretch(shrink)doublespread{} can correct problems many pages ahead, 
but none of these is likely to work on the first attempt.  (I have not 
even mentioned \looseness, which someone condemned about a year ago as 
useless. I could not live without it.) Of course, if there is no need 
for balanced spreads, or verso-recto baseline alignment, none of these 
refinements is needed, and these many passes through DVI can be avoided, 
but no press I work with would accept feathering or ragged bottom 
spreads unless I could show that there was ABSOLUTELY no alternative.  
So far, that (ragged bottom---NO feathering) has happened only once in 
twenty years of journal setting.

Pierre MacKay