Hi,
ext Matan Ziv-Av wrote:
>> Telnet isn't something that's either:
>> - Needed by the device itself
>> - An essential (needed in installing most of Debian packages
>> without them declaring a separate dependency)
>
> The same is true for chvt, netstat, uniq, and probably many others, yet
> they were included.
chvt: needed by device bootup scripts
uniq: essential (also POSIX standard utility)
netstat: legacy. we've considered removing it (along with some other
legacy non-essentials in Bysybox), but our own developers use
it often[1] and removing it from public releases doesn't really
save that much space
The reason why I talk about our own developers is that their situation
is very different from external application developers & (advanced)
users.
Our own developers may need to flash new releases many times
a day and need to debug situations where the new internal release
cannot even connect to network at bootup. I.e. installing additional
packages from repositories isn't really an option.
However, the published releases don't have anymore these kind of issues
and releases happen only couple of times per year, so for end-users
installing tools from a repository after downloading & flashing
a new release is completely valid option and a right thing to do.
> The decision to not include telnet was a bad
> decision. Admitting to mistakes and fixing them (even if only in
> Fremantle) is better than attacking someone who expects to have telnet
> on his internet tablet.
If my reply seemed like an attack, I apologize, it wasn't meant as such.
>> It's an (advanced) end-user tool and can be installed separately.
>>
>> Have you tested that Busybox telnet even works as well as real telnet?
>> With the past experience I have some doubts about the quality of misc
>> Busybox tools. :-)
>
> Yes, I have. And even if it does not, it surely does better than no
> telnet at all.
The problem of adding new things to Busybox is that:
- they then conflict with the real alternatives
- they increase our essential/base set (busybox is essential so
anything added to it becomes then an essential too)
That's not good from the package management point of view.
See e.g. bug:
https://bugs.maemo.org/show_bug.cgi?id=2896
This has been discussed quite extensively on this list and in
bugzilla. I hope we have a solution for this in Fremantle.
> I wonder what kind of QA did Nokia run on busybox's top or nslookup (for
> example) before deciding to include them?
Busybox top is (still[1]) crap and is problematic because it conflicts
with procps that provides much nicer top and also many nice other tools.
However, (some kind of) top being pre-installed on the device is
really a must for us & useful for all developers and many advanced
users.
nslookup removal has been considered but hasn't been done so far.
[1] Despite the multiple rewrites on the code I originally sent
to Busybox authors. :-)
>>> Life with the tablet could be a bit simpler if Nokia chose to include more
>>> stuff that is available in busybox (md5sum, cpio, hexdump, ...).
md5sum is an essential + there's a bug about it missing in
bugs.maemo.org. As a result it's included into Fremantle busybox
(along with quite many other utilities belonging to Debian essential
packages Busybox claims to be providing).
cpio & hexdump aren't essential (in the upstream compatibility
/ Debian packaging sense), so they can be installed separately.
"cpio" is in Debian in "cpio" and "hexdump" in "bsdmainutils"
package.
>> Some of them will be included into Fremantle; the ones that are in
>> Debian in packages that Busybox currently claims to provide and
>> conflicts with and which provide options compatible with GNU tools.
>> Other tools belong to somewhere else.
>
> telnet belongs on an "Interner Tablet", and claiming otherwise is akin
> to claiming that the WWW is the internet.
Device has application manager & apt-get. Claiming that in
an "Internet" device everything needs to be pre-installed into
the device itself is so mid-90's. :-)
- Eero
_______________________________________________
maemo-users mailing list
maemo-users@xxxxxxxxx
https://lists.maemo.org/mailman/listinfo/maemo-users
