Questions: why for pycrypto needed?

[James Blackwell]

On Fri, Jan 06, 2006 at 01:53:38PM -0600, John Arbash Meinel wrote:
> Alexander Belchenko wrote:
> > John Arbash Meinel пишет:
> > I probably miss something about bzr. On newly added page in wiki:
> > 
> >>> http://bazaar.canonical.com/BzrInstallation I read that bzr want
> >>> pycrypto package. Is there some new features is coming?
> >>
> >>
> >> Paramiko requires pycrypto. And bzr requires paramiko for sftp support.
> > 
> > 
> > I'm probably bore man. But pycrypto mentioned in top 3 must have things
> > for bzr. Paramiko mentioned later in 'strongly recommended' section. My
> > brain did not build bridge between paramiko and pycrypto. So, my first
> > thought was: 'wow, to bzr added new cool features?'
> > 
> > -- 
> > Alexander
> 
> Well, James Blackwell was the one who wrote the page. So before I fix it
> up, I'd like to know what he was thinking.
> Basically, I would move it from Required into Strongly Recommended.

I hear you guys. I think your concern lays in two areas; that it express
more dependancies that we need and that it complicates install more than
is necessary. I respect these motivations. 


As far as I can tell cygwin still relies on pycrypto being there when
using sftp. All of the installs I've supervised, including the one at the
beginning of this week, ultimately failed during use when pycrypto was not
installed.

If this is in error, then please remove that as quick as possible. If it
isn't in error, can it be made into one?  I say this because pycrypto
seems to not be in Cygwin for export restrictions.

> Also, cElementTree is only Strongly Recommended. Though it is very
> strongly recommended for performance reasons. (10x faster for reading
> XML than elementtree).

I think useless if you don't is practically a requirement. In my
experience most users are human thusly inherently lazy. I believe, with
experience, that users underrate requirements. Strongly recommended turns
into "install this first if it breaks". Recommended turns into "stuff I'll
play with in six months from now" and suggested is "stuff I'll never need"

Many windows users are not comfortable with extentensive command line
work, though they seem happy to keep a shell open just to run bzr
commands. Among the set of those that are familiar with cygwin, fewer
still are familiar with python installs.

It seems more productive to me if users can take their medice one time so
that they can use bzr. I see these possibilities:

 * They mistakenly believe that the behaviour that they experience without
   celementree is standard and throw it away as being unrunnably slow.
   They either:
     * Slog it through and suffer unnecessarily

     * Throw it away for something else

 * They start hitting mailing lists, asking other users and reading more
   docs to see how they can find out how to make the speed better. This
   cascades into three possibilities:

     * Swear up and down that they've got to do all of that mystical
       python crap.. again.. for a tool that has already let them down
       when it comes to performance and walk away

     * Shrug and install celementree anyways.

Out of of the four possibilities I see, half of them are failures and one
is a disservice. A 75% failure-to-satisfy rate is a steep cost for praying
at the alter of strict honesty -- especially when this module is already
installed on half of the systems out there and trival to install on the
other half (even in cygwin)


> 
> John
> =:->
> 



-- 
James Blackwell's home :  http://jblack.linuxguru.net
Gnupg 06357400   F-print AAE4 8C76 58DA 5902 761D  247A 8A55 DA73 0635 7400
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060106/c4bcfd02/attachment.pgp