[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Pnet-developers] An automatic cvs pull patch for ml-pnet

From: Gopal V
Subject: Re: [Pnet-developers] An automatic cvs pull patch for ml-pnet
Date: Sun, 1 Jun 2003 11:29:32 +0530
User-agent: Mutt/1.2.5i

If memory serves me right, Rhys Weatherley wrote:
> I don't really like this.  Doing a CVS pull on every make is very 
> inefficient, and blowing away the entire tree and restarting on a 
> make distclean will be incredibly wasteful of bandwidth.  

Nobody is more concerned about bandwidth than me ... this is for the 
newbie developer who click-downloads ml-pnet-0.0.2.tar.gz from Southern 
Storm and is trying to build it ...

I think a good readme item about the PULLCVS mode might help many who 
have no idea or experience using CVS and don't want to be bothered by
it. Especially important in this category are the various RPM vendors
building pnet.

> (I do a "make distclean" several times a day, for example).

Like I said , people who do "make distclean" on cvs pulled stuff deserve
the effect they get :-) .. The pulled files are generated data during the
build , they are to be distcleaned !

People who don't use PULLCVS will have the build as usual .. which I 
consider is the ideal scenario.

In short, you are not among the target audience :-)

> build must be kept in two places: "ml_pnet/" and 
> "ml_pnet/class/".  It would be better to pull the entire "mcs" 
> subtree.

Bandwidth ... why download the 14 MB of corlib & System when you don't need
them at all ?. (we need a ~600 byte file from corlib, but that's no real 
reason to pull 14 MB of it)

mcs subtree== 45 Mb
sources.cvs== 4.3 Mb

See why I picked the latter option ?

Also I did try many ways of obtaining the files to download , the 
best option so far was 

find tools class | grep "\.build" | sed "s/\/[^/]*$"

But that's putting some extra pressure on the guy naming the directories
to be identical to the CVS paths.

> Not in this form.  It's a little too clever.  A better README should suffice.

I don't want to tell a prospective developer to read the CVS instructions
of another project , download it and then compile this module ... It just
doesn't make sense . To be able to build this without having to poke through
their website for CVS instructions is a good thing for pnet .

If you put a little too much strain on most people , they will just give 
up the build ... the "clever" part is what makes this important :-) ...
Let the machine handle the hard-coded stuff ...

This program is simply an attempt to make this module at least a small 
step towards self-containment. It's an alternative , and not a replacement
for the regular --with-....sources option.

The difference between insanity and genius is measured by success

reply via email to

[Prev in Thread] Current Thread [Next in Thread]