On Jul 24, 2016 6:04 PM, "Roberto Verzola" <[email protected]> wrote:
>
> > there was never a  notion of a partially delivered message, nor one that
> > could be resumed.
>
> I find it hard to believe this one. The Internet from the beginning has
always been about partially delivered files.

Not only is there no "single" coherent Internet design (the network
protocols comprising different parts of what we recognize as tne Internet
often have conflicting and overlapping goals and responsiblities) but mail
- as in the intrahost messaging system - preceded the Internet. It was
barely syntax sugar over appending text to a shared file - the existence of
which meant that you were already sharing files reliably. So message resume
was just not factored into the design.

And to add a "simple" feature as resume takes more work than you might
think. it would require, for instance, indexing all the incompatible
mailbox formats with reliable, non-repeatable ids. SMTP - client-side -
would have to be overhauled to reliably send message sizes or wait for
application-layer ACKs to message chunks (messages arent chunked right now)
in a time when we can barely trust outlook not to mangle MIME. And then
theres defining what to do in the case of multiple redundent relay servers
- who stores the incomplete message? the server that happened to be
roundrobined that minute or an authoritative sender or the ultimate
receiver? Or multiple levels of smarthosts / relay servers which are all
user-transparent... what happens when transmission needs to be resumed at
an intermediary? Can (or should) users read partial messages? What happens
if a partial message is updated while reading? How do MIME attachments get
chunked? Is there a checksum mechanism? How do you deal with brute force
message injections (smtp is largely unauthenticated at the receiving end -
hence why spam works)? etc...

No doubt a decent programmer can make clean room designs to the questions
above - which is probably why we have a hundred different messaging
protocols - but thats because we have the luxury of thinking about the
problem after years of real world testing. They simply were never factored
into email protocols and its not easy to add them in now.

>
> Greetings,
>
> Obet
>
> On Sat, 23 Jul 2016 15:26:40 +0800
> Mark David Dumlao <[email protected]> wrote:
>
> > On Jul 21, 2016 18:35, "Roberto Verzola" <[email protected]> wrote:
> > >
> > > > Why not just upload the file somewhere that supports upload resume
then
> > > > just email the URL. Most FTP servers, and cloud storage, services
like
> > > > Dropbox or Google Drive, supports resuming of interrupted transfers.
> > >
> > > Maybe, but that is too much work, especially if the problem occurs
> > regularly because you have a slow connection. Email with attachments is
> > such a universal Internet event that I expect by this time that such
> > 30-year (at least!) old technology as resumed file transfers would have
> > found their way and become part (transparent and automatic) of the
process.
> > Or am I expecting too much?
> >
> > Yeah you are.
> >
> > Regardless of what theyre used for today, some network protocols were
> > designed for something else entirely and enhancements to those protocols
> > tended to stick within the limitations of those designs.
> >
> > Email historically came from the mail systems from inside a multi-user
> > machine, which worked under the assumption that mailbox files were
written
> > atomically and quickly with relatively short, text-based messages. Thus
> > there was never a  notion of a partially delivered message, nor one that
> > could be resumed.
> >
> > Furthermore more reliable file transfer was always available throughout
the
> > history of mail protocols. In intra-host mail, the fact that you shared
a
> > filesystem meant that there was no point to duplicating the files. In
> > host-to-host mail, other methods for serving and transferring files were
> > always around. And many mailservers simply had attachment size limits
and
> > the like so large attachments never got a serious consideration. (it
would
> > be trivial to dos a server or user that arbitrarily accepted large
files,
> > like email does)
> >
> > So tldr: the reality is that email was never intended for heavy file
> > transfers and that is unlikely to change any time in the near future.
Mail
> > has historically always relied on other file sharing methods.
>
>
> --
> Roberto Verzola <[email protected]>
> _________________________________________________
> Philippine Linux Users' Group (PLUG) Mailing List
> http://lists.linux.org.ph/mailman/listinfo/plug
> Searchable Archives: http://archives.free.net.ph
_________________________________________________
Philippine Linux Users' Group (PLUG) Mailing List
http://lists.linux.org.ph/mailman/listinfo/plug
Searchable Archives: http://archives.free.net.ph

Reply via email to