On Sat, 10 Sep 2022 21:04:54 +0300
unix <u...@disroot.org> wrote:
> Hello. My reasons for this proposition:
> 1. The user will be able to test basic websites without installing
> anything.
> 2. The user will be able to read an incredibly useful official
> FAQ, with no external devices involved.
> 3. The user will be able to manage mailing list
> membership via the web interface.
> 4. Using ftp(1) and reading pure HTML is inconvenient.
> 5. The browser (Lynx) was already included. It was removed due to
> concerns about code quality, licensing, and support for insecure
> protocols.
> You could say that you don't need a browser installed by default if
> you have a network connection and can install the package anyway.
> Still, am I the only one who feels like it's pretty much the only
> thing missing in the base for a comfortable day to day desktop usage?
> So, if we include a browser, which one?
> I could only find two options. The rest use
> unacceptable licenses or are way too big and scary.
> w3m:
> - terminal only.
> - No CSS or Javascript support. CSS can be used for tracking. CSS3 +
> HTML5 is Turing complete. I don't need to convince you that
> Javascript is a security nightmare.
> Sources:
> https://lemire.me/blog/2011/03/08/breaking-news-htmlcss-is-turing-complete/
> https://www.templarbit.com/blog/2018/03/20/tracking-users-with-css/
> - No support for obscure protocols. Just HTTP/S and FTP.
> - MIT license. Seems to align well with our copyright policy.
> (For those who don't know, it's an equivalent to ISC and BSD
> 2-clause. Xorg also uses MIT).
> - Can also be used as a pager/e-reader with bookmark
> functionality. Two birds with one stone? Anybody feel like replacing
> less? Joke.
> - Optional image support (we shouldn't use it for security concerns,
> obviously).
> - 70k of C + boehm garbage collector, another 40k.
> Current source is maintained by a a Debian developer.
> https://tracker.debian.org/pkg/w3m
> w3l, fork of w3m:
> - Removed some features that can be provided by other tools (e.g. tab
> functionality).
> - Completely removed image support. The keybinding to open an image in
> an external viewer is enough.
> - For some reason uses curl. I guess, it can
> use ftp(1) instead.
> - Minimal reliance on the garbage collector. I suspect that it can be
> completely removed.
> - 30k of C + 40k of C in boehm gc.
> Source: https://sr.ht/~vdupras/w3l/
> If you know about any other options, I will be interested in
> discussing them.
>
Several days reading the archives and sources, here are my thoughts.
w3m consists of two parts:
1. html to text converter
2. a pager.
I don't see a point in including a pager since we already have less, vi
and mg.
I don't see a point in doing an html to text converter in
C. Perl serves this purpose quite well since text manipulation in Perl
is awesome. One module for this purpose is already in ports.
https://metacpan.org/pod/HTML::FormatText
But then, after thinking a bit more, is there a
point? Is there anybody who needs this? The average OpenBSD user
is capable of reading. If something that he
needs to read is missing, we should just add it to the usual locations
in plain text. HTML conversion is not needed for 99% of cases.
I should do something else.
I posted this for anybody who thinks that OpenBSD needs a web browser.
Not everybody needs to read the web, and if you do, it doesn't mean that
the tools for this should be included in the base.
