[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: HTML HOWTOs and Re: CD edition
On May 5, 12:26am, David Lawyer wrote:
> Subject: Re: HTML HOWTOs and Re: CD edition
> > On Apr 30, 11:03pm, David Lawyer wrote:
> > > Subject: HTML HOWTOs and Re: CD edition
> > ...
> > Note that a gzipped tar file is also created which is a package
> > of all the single-file HTML docs.
> > The question is "should the files themselves and/or the tar
> > packages be mirrored...."?
> With data compression used on most modems, tar doesn't save much
> bandwidth anymore. Since disk space is cheap, the major cost would be
> the bandwidth costs for updating the mirror sites. But people who
> download HOWTOs from their local mirror site (instead of from Metalab)
> would conserve bandwidth. So I think it's a good idea. Since
> packages of all the HOWTOs come with various distributions of Linux I
> think we should supply individual HOWTOs for downloading from the
> mirror sites. For those who don't know, we now don't supply any
> HOWTOs for downloading from the mirror sites. If we do this we would
> be supplying single-file HTMLs from the mirrors.
Sure we do. Currently the tar package for the HTML HOWTOs,
(the multi-file HOWTOs) are available on the mirror sites and
do not require being pulled from our host machine.
> Should they be tarred? If not, then they could be used for on-line
> browsing and anyone who wanted a copy would use the browser to save
The tar packages are nice for those that want to grab
the entire set quickly/easily (for offline browsing or
> But this wastes bandwidth for those that only read a small part
> of one. One solution is to have 2 ways to read on-line: HTML-split
> (most efficient --what we do now) and HTML-single (can easily save it
> with the browser). To encourage efficiency, I would only say "read
> HOWTOs online" for the split ones.
> Another issue is that of small HOWTOs. There isn't much point in
> splitting a 10k HOWTO up into 10@ 1k chunks. If someone reads even
> half of this small split-HOWTO they probably use more bandwidth in
> overhead than they save by transferring less text-data. For this case
> they are better off reading the single-HOWTO. But how to easily
> implement this?
Not easy. The split is based on document structure (when
processed). I suppose we could do something like -
if the file size of the single-file HTML variant is less
then X bytes, replace the multi-file HOWTO of the same name.
I'm more inclined to remain consistent with the presentation,
however - even if that means 1K chunks (just my opinion).
> One could also argue in favor of plain text which is what I use but
> with possible future interlinking between HOWTOs, I would vote for
> Since we have been giving people only the split html-howtos for
> downloading, I guess it's OK to keep doing this. But now we give them
> a choice. Most people who understand this option should take the
> single html-HOWTOs since they are easier to search when using a
> browser. Also there are a lot fewer files to find with ls (or
> locate). Thus the distribution of the split ones should be of little
> interest. Except that the split ones are used for on-line reading to
> conserve bandwidth and are sent to all mirror sites for this purpose.
Ok, so based on those comments, I believe that the tar package(s)
for both the split AND single-file HOWTOs should be made available
from the mirrors.
What about the single-file files themselves (not just the tar file)?
I think they should probably remain accessible only off the host
machine (and not mirrored).
> From the way it looks now in docs.html#howto most people might just
> choose "HTML" and not understand what is meant by "single-file". Thus
> I think that the split ones need to be called "HTML - split" with the
> "single-file" ones coming first. There are probably a few sites that
> have added useful material to the indexes that need the split ones
> (although they could generate them from the .sgml files). Eventually,
> "HTML - single file" could be changed to just HTML since they would be
> used much more than the split ones.
Yes, but the split ones are still the primary browseable
docs, right? That's what you said earlier in your msg (if
I read it correctly).
So, in summary (correct this if I misunderstood) :
* mirror the split HOWTOs for (primary) browsing -- DONE
* mirror the tar files containing the split HOWTOs -- DONE
* keep the single-file HOWTOs on the host only -- DONE
* mirror the tar files containing the single-file HOWTOs -- OPEN
Greg Ferguson - s/w engr / mtlhd | firstname.lastname@example.org
SGI Tech Pubs - http://techpubs.sgi.com |
Linux Doc Project - http://www.linuxdoc.org |
To UNSUBSCRIBE, email to email@example.com
with a subject of "unsubscribe". Trouble? Contact firstname.lastname@example.org