Re: Rewrite cvsup & portupgrade in C

On 07/07/04 22:26, Steve Kargl wrote:
> On Wed, Jul 07, 2004 at 10:56:17PM -0400, Wesley Morgan wrote:
>> On Wed, 7 Jul 2004, Steve Kargl wrote:
>>> My initial argument does not invlove the language.  I don't care
>>> about the language.  My argument is that neither cvsup nor csup
>>> belong in the base system.  Both utilities can be installed from
>>> ports.  If you're going to import csup, then I hope csup goes
>>> through a security audit and you define a NO_CSUP make.conf
>>> variable.
>> 
>> Does "csup" belong in the tree more, or less than cvs?
> 
> csup does not belong in the base system.  cvs belongs in the base
> syste
> 
>> A program that I have used maybe a handful of times because cvsup 
>> is so much better for my purposes is rebuilt with every make world 
>> unless you explicitly disable it (which I would argue that most 
>> people do not).
> 
> cvsup is built with every make world?

I think you misunderstand: He was questioning why *cvs* is in the base 
system and is built with every make world when many users opt for the 
more advanced cvsup instead (I used cvsup exclusively for my first 
couple years with FreeBSD).  I can see his point, but I think it's a 
good idea to include in the base system the tools used to contribute to 
FreeBSD.

I don't think it matters if c(v)sup is in the base system.  Even if it's 
not, the installer could install the package by default similar to the 
way it handles perl.  I think c(v)sup should be installed by default, 
but I couldn't care less whether or not it's in the base system.

>> It has had security holes and other issues.
> 
> And csup won't have any issues?

It's a moot point at this stage anyway.  I don't think you could 
convince many people that adding csup to the base system this late 
before 5.3 is a good idea.  If csup proves itself as a port (which I 
have no doubt that it will), perhaps adding it to the base system can be 
revisited for 6.x.

>> Why is this in the base system at all?  Simply so developers can
>> make commits from a fresh install?
> 
> Rhetoric, but correct.
> 
> You do realize that you can use cvs to retrieve the FreeBSD sources 
> via anoncvs.  So, cvsup isn't even needed.

There's a reason why we have cvsup.  It's actually faster for me to 
mirror the entire cvs repository locally using cvsup (see 
/usr/share/examples/cvsup/cvs-supfile) and 'cvs update' from there than 
to use anoncvs.  The speed of cvsup and doing the 'cvs update' locally 
more than make up for the extra steps involved.  Granted, I have a 
decently fast machine with lots of RAM, but anoncvs is still *much* 
slower than cvsup (sometimes by several orders of magnitude).

Jon