==========Handling FTP/browser downloaded files
The GPL utility untgz.exe is able to do untarring from the command line in
a DOS[box] and works well with ftp. It is equivalent to unix/linux tar xvf
plus either gunzip or unzip/pkzip25 if needed. In my experience it is fast
and accurate to the extent that either MS or linux OS can be used for file
acquisition for linux use.
Author Tillmann Steinbrecher keeps this well-developed compact utility up to
date and makes it available on simtel,
ftp.elf.stuba.sk/pub/pc/pack/untgz???.zip
etc.
You can tell if files are tarred without depending on the filename: Seen* by
looking at unzipped files --whether ASCII or binary or mixed --, [unzipped]
tar composites have header blocks (mostly @-signs with the name of the
component file) for each component file. (*Literal views from, for
instance, a DOS[box] running Buerg's list.com or Shawn Hargreaves' editor
fed.exe or less.exe (see www.delorie.com for actual C/unix-utilities that
work using x86-ported gcc)).
===========Download/upload control control and system design
Thanks for the data on browser downloads and control, guys. Since we
develop thru the intermediary of down/uploads it's a critical real life,
real time! communication control concern.
The download accuracy problems we're finding might be soluble by fixing the
code either with a form as noted in the prior message (below) or a
command-line option.
<<there's no way to select whether Netscape then does a
binary or an ascii transfer.> Andrew Cannon <[EMAIL PROTECTED]> >
<<Yes, netscape adds something. When I down loaded a disk image
of the linux router project it was too big to fit on a floppy
by about 6k bytes. I downloaded again but with IE and it was
the correct size. Of course I could have just used ftp. Lou>
<[EMAIL PROTECTED]> >
<<It seems to me that the most obvious solution would be for the 'download'
popup box to have a field for selecting ascii or binary transfer, with the
default being setup each time according to the current rules. Then you
could see what it was going to do and would have the opportunity to change it.>
Andrew Cannon <[EMAIL PROTECTED]> >
There are command-line options for browsers too, though that's too
straightforward -- well, maybe not. [[Also see Steve's info below.]]
The Netscape browser sourcecode is out there and being worked on (so I'm
forwarding this to the responsible group, Mozilla). Netscape's hopes of
prevailing via the freeware method led to their sourcecode for Netscape
release under Netscape Public License in April.
Downloadable reworked versions are at ftp.mozilla.org/pub/mozilla/sources/
with discussion at www.mozilla.org and a newsgroup. The original three
versions have been combined into two now, Windows+unix (18.7MB) with
different files only where needed, MacPPC (19.2MB) bigger in spite of being
single platform, with added compile tools (1.1MB) for each of the 3 platforms.
Since the release (1998-04-28), Mozilla tech honcho Brendan Eich reports
Oct. 26 that things are moving along: To date the mail and news reader code
has not shown up, but new ones are being written. Autoconf will be std. for
compiles, with modularity through XP-COM and compatibility across modules
thru XPFE. (Because the original FE lacks the kind of hooks emacs elisp
provides.) There are 2 domains of code, XP cross-platform and FrontEnd for
specific display platforms. There's to be an embedded HTML composer/editor.
Open code is taking precedence, as in moves to Berkeley DB for the mailer
and anticipated replacement of unix/linux X-windows Motif windows widgetry
with GTK+ . They hope autoconf will work even in W32 using cygwin.
Makes an interesting development context comparison to look at
Browser as a HIGHER level "PLUGIN" with PLUGINS (but IE is largely
HIDDEN as an INNER level) talking across machines
to
RTLinux as a LOWER level "ADD-ON" with INNER level hardware controls
[with potential for multiple-CPUs]!
SC
========================
<<To: [EMAIL PROTECTED]
From: Andrew Cannon <[EMAIL PROTECTED]>
Subject: Re: [rtl] Source for un-archived source on Web
Cc: [EMAIL PROTECTED]
...
>Our experiments indicate that the way Netscape handles file downloads
depends
>on the link type and the mime header (if any), and the file extension.
>
>If the link is ftp:// then it seems to always use binary mode.
>
>If it is http://, and there is a mime type header (Content-type:
>application/octet-stream\n\n) then it uses this information for the download
>method. Application/octet-stream is always binary mode, as far as we can
>tell. Web sites should always wrap files with mime headers, to prevent
>unintended results. This can be done either with cgi scripts or by
editing the
>file itself. Obvously for archives, cgi must be used.
>
>Otherwise the file extension is used.
>
>There are occasional exceptions we have found, notably bugs in audio file
>players (maybe fixed by now). In these cases the mime header was ignored
and
>the file extension was used to determine what to do with it.
>
>The association with the file extension and how to treat the file is set
under
>edit->preferences ->navigator->applications, and then edit. Maybe you
need to
>set the file extension you were downloading. The mime type should be
>"application/octet-stream", and set it to always save to disk
>
>In general MSIE when you get into technical areas does things wrong in more
>cases than Navigator. For example, MSIE does not handle dynamic document
mode
>(multi-document mime types), ie, server push. Also, MSIE will not allow ftp
>uploads, or at least, we haven't figured out how to do it.
>
>But both companies should take their code out back and shoot it; they are
both
>way too buggy, leak core, and crash way too often.
>
>/sg
Dr. S. Candelaria de Ram
Cognition & Communication
--- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
----
For more information on Real-Time Linux see:
http://www.rtlinux.org/~rtlinux/
