Re: Web versions

[Jean Louis]

* Jacob Bachmeyer <jcb62281@gmail.com> [2021-03-17 05:09]:
> Jean Louis wrote:
> > * Jacob Bachmeyer <jcb62281@gmail.com> [2021-03-16 10:30]:
> > >    3.  Web apps stored on "the cloud" are bad because they often do not
> > > respect the user's freedoms, as even if the software is under Free license
> > > terms, technical issues can make running a modified version difficult or
> > > impossible.
> > 
> > Just because there is possibility of abuse one shall reject the
> > technological opportunity?!
> 
> That is ridiculous, but we should still take steps to mitigate possibilities
> for abuses.  After all, we have the GPL to mitigate the abuse of "walking
> off" with a copy of a Free program and making a proprietary derivative, and
> GPL3 was introduced to mitigate the abuse of Tivoisation.

Exactly. We can mitigate it:

- plugins for safety of work with Webassembly for now do not
  exist. There is this Webassembly detector:
  
https://addons.mozilla.org/en-US/firefox/addon/webassembly-detector/?utm_source=addons.mozilla.org&utm_medium=referral&utm_content=search
  and there is possibility to disable it completely.

  In my opinion we shall have a plugin that asks user if to run it or
  not and start making list of safe websites with Webassembly.

> >  Or maybe you wanted to define "GNU operating
> > system" as only those software packages developed by GNU, but not
> > those software packages delivered with the GNU operating systems like
> > Parabola in my case?
> 
> The original request that started this discussion was a suggestion to port
> all of the software developed by GNU to WebAssembly to run in browsers.
> There are some packages, such as coreutils, for which that is obviously
> nonsensical.  This does not mean that we could not support WebAssembly as a
> general compilation target, or that we could not build effectively a
> browser-based HURD port, (and HURD's architecture fits such an environment
> fairly well, treating the browser itself as a Mach-analogue) but generally
> relying on browsers is dangerously close to SaaSS.

In one of the referenced hyperlinks I have shown that somebody has
already started with GCC, binutils in Webassembly. I do not see why
coreutils do not make sense. Why you think so? How to handle all the
scripts when compiling software?

Maybe review the fact that Webassembly need not run in browser, it can
run standalone, that makes such programs cross platform, thus
useful. Coreutils are then easily deployed on various systems.

Binutils, GDB is already there:
https://github.com/pipcet/binutils-gdb

> but generally relying on browsers is dangerously close to SaaSS.

It is generalization. If you know a bug, why not report specific bug?

Web browser with Webassembly is not anymore a web browser only, just
as Emacs is not just editor.

> The original request was for GNU packages to be offered as SaaSS.

Making GNU packages run in Webassembly does not make it automatically
hosted or served by third party. There is difference between SaaSS and
software alone.

As software it is useful to have possibility to run GNU, various other
free software, including Emacs in Webassembly, or complete OS-es.

If some company or party will run any software as SaaSS that is their
choice. Serving it or selling it in that way is separate issue of
having software for Webassembly.

Having software means: user can run it from their own computer after
downloading and making informed decision as Alfred said. Users can
host it on local area network. But need not. Users can host it on
their websites, LAN or similar. But need not to, they can execute it
even from USB. Thus usefulness of having software that runs on every
OS is separate from SaaSS.

Jean