[Info-vax] Re; Spiralog, RMS Journaling (was Re: FREESPADRIFT)

[Jan-Erik Soderholm]

Den 2016-06-24 kl. 04:15, skrev David Froble:

>
> But, to get serious for a moment, (yes, I can do that now and then), in
> practice perhaps it's not such a big issue.  For example, I'm not
> everybody, but, I do not send large pieces of data.  I've never sent
> anything that could not easily be read into memory prior to transmission.
> Got to ask, in general, how often is rather large text files transmitted to
> / from a web server?

Any fairly modern web page can have 100s of different objects. Large text
files might not be that common, but there are usualy a lot of images
for buttons, icons and general page contents.

If we look at WASD that I know about, it maintaines an internal file-cache
where both the HTTP headers (which includes the byte-size) and the actual
content is saved in a format that is fast to retrive and send. So instead
of opening the soure file, a cached copy is used.

How this cache works is clearly described in chap "7.2 Caching" in:

http://wasd.vsm.com.au/wasd_root/doc/features/wasd_features.pdf

>
> If I FTP a data file, it's a binary transmission, and the byte count is
> simply filesize in blocks * 512.  I have moved large data files.

FTP has nothing to do with a web server. And a binary file does not
have to be a multiple of 512, of course. It can have any byte count.

>
> Ok, in custom communications I've implemented, the byte count goes first.

And so it does in the HTTP protocol.