[Vorbis-dev] Again: Next libvorbis release?

[Max Horn]

Am 05.03.2009 um 23:27 schrieb Ralph Giles:

> On Thu, Mar 5, 2009 at 7:39 PM, Max Horn <max at quendi.de> wrote:
>
>> I am done with the initial LaTeX conversion
>
> Wow, thanks for doing that!
>
> Doesn't look bad. HTML output from LaTeX has come a long way. Still
> get nonsense anchor names, but the docbook stylesheets did that too.
>
> pdflatex worked fine for me on both macos and linux.
>
> Issues I noticed in a quick look through:
>
> * I'm surprised it can't import xifish.pdf instead of the png version?

Well, pdflatex can use xifish.pdf. But it doesn't because I add this  
line:
   \DeclareGraphicsExtensions{.png}
Now, you can get it to prefer the PDF by changing it to:
   \DeclareGraphicsExtensions{.pdf,.png}

Why didn't I do that then? Well, in an attempt to get htlatex use the  
PNG version and pdflatex the PDF version. In the final version, I'll  
change it to executing "  \DeclareGraphicsExtensions{.png}" *only*  
when using htlatex -- this is not difficult, just not yet done.


>
> * Chapter 8 is labeled vorbis-spec-residue instead of
> vorbis:spec:residue, so references to it are not resolved.

To think I used to believe that I know how to use regex search and  
replace... :). Fixed.

> * The last paragraph of lookup1_values in chapter 9 has a continued
> \varname{} which should be split in three.

Fixed. Note that I converted it straight from the Docbook, where it is  
the same way.

> * A number of the Verbatim pseudocode sections have too long a line
> length for the pdf pages.

Yup. Just as in the Docbook files. I made a point out of *not*  
touching that. Definitely not great, but at least it is not worse than  
before, and no regressions due to me mixing up something in the  
formatting.


Another thing some people might find annoying are the boxes that  
surround all hyperlinks in the PDF. I can turn this off, or  
alternatively, make the link *text* colored. The colors themselves are  
also customizable. Just tell me what you prefer.


> On the last one, originally the spec was written directly in html
> (circa 2001) and all the pseudocode was like this. Before stylesheets,
> it was a reasonable choice. When I did the docbook conversion I
> started converting everything to numbered lists, but wasn't entirely
> happy with the results; especially as later the pseudocode gets much
> more code-like, with flow control, but still occasionally wants
> reference to line numbers. All of which is to say, currently it's not
> consistent, and you should feel free to do any third thing that works
> better.


Eventually, I think the whole pseudocode should be changed to either  
use the listings package or algoritmicx or something like that. For  
both, the result will look much better, and will get nice bonuses like  
keyword formatting, easy indention change, line breaking etc..

I could take care of this, too, but I'd prefer to do this in a second  
step -- I'll *only* do this if I know it is not going to get thrown  
away. I.e. I'll wait to see if my work so far (after I cleaned up all  
buglets you still uncover) is accepted at all. For the same reason, I  
haven't bothered integrating this into your build system. I already  
spent considerable time on this, and before I spend much more, I would  
like to know whether it is thumbs up or down on this :).


> Likewise with the '[long_name]' notation for vectors.

Aye, definitely should be changed. But again, I thought it best to  
first make the huge format change, and verify that there are no  
regressions in the content. Then, once there is a stable base, one can  
work on improving that.


This leaves me with a few questions:

1) Can anybody comment on the wrong formulas I pointed out in my  
previous post?

2) Assuming I fix the points Ralph mentioned above (and whatever other  
stuff you guys through at me, assuming it is reasonable), who is  
responsible for deciding whether this get

3) So... let's assume for a moment you guys accept this, and thus end  
up with properly building specs/ docs again...


   When is the next libvorbis release?
   What is blocking it?
   Any way I can help?  :-)



Cheers,
Max