[arachne] Re: unifying the sources

["Udo Kuhnt" <048321887-0001@xxxxxxxxxxx>]

Arachne at FreeLists---The Arachne Fan Club!

Hi Ray,

>> I think Jason is absolutely right, and if you remember, this is just
>> what I suggested to you. If you are really interested in getting us to
>> use your code, then breaking it up into small sets of patches that we
>> can integrate in ours would be your best chance to achieve this.

> I wish it were possible, but it is not. The entire structure has been
> changed, so a series of cut-and-patch simply can't work.

that's nothing that DIFF and PATCH could not handle.

> Besides,
> why bother with that anyway? I have yet to understand why you would
> not want to adopt my sources, since I have all current work except
> IRC and news. Even if it was only a tiny bit better, it is still
> better!

There is a good reason for this, and that is continuity. Going back and
forth between two sets of sources to keep track of changes in the
behaviour is hard enough if they are just different versions of the
same source tree. Chasing bugs across different trees is a nightmare,
and would put an unnecessary strain on a project that is already short
on both developers and spare time.

To use your argument against you, the time necessary to merge your
sources with ours is proportional to the time it would cost us to adopt
to yours. If you would narrow the gap, it would make switching to your
code easier; that was just was I had suggested.

> That's a subjective judgement, so it can't be debated, however there
> are some hard facts. But first, if your last try with my stuff
> was back in the 'rcore.exe' days, things have progressed greatly
> since then. As Eric will tell you, 'rcore' was the time of the
> greatest difficulties. Anyway here are some hard data comparing
> my latest cores with J3:

> As you can see, it is the opinion of the compiler that my sources
> are about twice as easy to read. And the 'sloc' program thinks they
> are ca. 18,000 hours of programmer time more simple.

IMO what you say about sloc is just as subjective as it can come. If you
want to talk about hard facts, then show us a minimal diff patch to
convert Joe's version to yours. Then we can see how different it really
is.

> Yes, there are some changes in data structures, but I have made
> *more* things static. My DGROUP was getting so empty that I've
> moved dynamic and FAR data in there to fill it up. That's the biggest
> reason for my speed improvement.

If that is true, we would get into trouble trying to make a DPMI version
from your source. Like I said back then, the DGROUP of the code in
Real Mode cannot be compared to the one of the DPMI version, because the
latter encompasses the whole stack segment. The trick to switch to a
different stack segemnt simply does not work with the DOS extended
version. For this reason, you would be well-advised to undo these
changes and make as much code FAR as possible if you hope to ever
compile it with the extender.

> Anyway, as I mentioned, my stuff is now being made into a proper distro,
> let's see what the users think of it. If it is rejected, then your
> possition 'wins' so to speak, but perhaps my core will become the
> choice of the users, in which case it is then the 'standard' and

I will at least give it a try and see if it works.

> I hope you and the others will respect the users decision.

We will see.

Regards,

Udo

-- The DR-DOS/OpenDOS Enhancement Project - http://www.drdosprojects.de

-- This mail was written by a user of The Arachne Browser - http://arachne.cz/
                  Arachne at FreeLists                  
-- Arachne, The Premier GPL Web Browser/Suite for DOS --

Other related posts:

[arachne] unifying the sources

[arachne] Re: unifying the sources