Compile times

Bruce Dubbs bdubbs at swbell.net
Thu Jul 26 05:08:58 PDT 2001


Mark Hymers wrote:

>On Thu, 26, Jul, 2001 at 12:03:24AM -0500, Bruce Dubbs wrote:
>
>>>OK.  There's been no opposition.
>>>
>>>Compile times are going :-)
>>>
>>>Mark
>>>
>>According to my system, you posted the first message at 05:16 and this 
>>message at 15:00.  10 Hours.  Please give us a chance to reply before 
>>implementing a policy.  At least 24 hours.  48 hours would be better.  
>>
>>I don't know where you are based, but I am in Texas and sometimes can't 
>>get to my system until late.
>>
>Sorry about that.  I hope that you understand that because the book is
>just starting, I'm anxious to try and get it moving quickly before
>inertia is lost.  Out of interest, I'm based in the Newcastle; in the
>north of England.  If you really want compile times to go back in then
>I'm still willing to listen but again my problem is that I probably
>won't have time to compile each package under similar conditions (i.e.
>with X closed - at at console prompt) and I'm not totally convinced it's
>necessary anyway.  If you look at the relative size of the Estimated
>Disk Space entry, you should be able to gauge the length of the
>compilation time from there (i.e. 1MB disk space required --> 30 secs,
>while 1GB disk space required --> Mozilla^H^H^H^H^H^H^H sorry - 1.5
>hours).
>
>Mark 
>

As I said on my earlier post on this subject, We don't need a lot in 
this area.  Something on the order of:
Approximate compile time: 24 minutes on  a P3 800MHz ssytem.

Where the variables are the time, type, and speed, of the sytem.  Of 
course there are other variables like RAM, disck speed, etc, but I'm not 
advocating a complete system descriprion.

Certainly I don't expect you to time everything in the book.  Each 
contribution can give this data.  Its really as simple as:
time make

I'm sure you will have losts of volunteers.

The relative size of the package size is an indirect measure.  I just 
feel the time on a specificde system is more direct and useful for the 
reader.  See my KDE hint for an example.

   -- Bruce





More information about the blfs-book mailing list