[illumos-Developer] webrev: printf

[Joerg Schilling]

"Garrett D'Amore" <garrett at damore.org> wrote:

> On Thu, 2010-10-14 at 21:17 +0200, Joerg Schilling wrote:
> > Something that shoule be detected by lint. Does this happen?
>
> No, it had passed lint.

Well, it is of the same kind of problems that AT&T had in every command source 
in 1990, due to the fact that SVr3 did map a 0 to address NULL for every 
process. As Svr4 is based on SunOS-4.0, these programs all dumped core when 
called without args ;-)

> > My impression is that there have been codereviwes for Illumos that were 
> > not done by people who fully understand the related code. I would not only like 
> > to see code reviews but wualified reviews.
>
> What kinds of qualifications would you like?

This depends on the code that is going to be integrated. I know which code I 
understand (see below for a list) and in case I get the impression that Illumos 
really becomes a collaborative community project, then I would take some time 
to review code I have the needed skill for. But this is a mutual process. You 
need to value other people's work in order to get your code valued by others.

> I refused to subject all integrations to some committee of known ksh93
> bigots.

If there are bigots, there is no need to name them this way. Just treat them 
the same way as everybody else and they should disappear....


> My changes here were simple enough to understand, *and* I got the review
> and approval of several different talented engineers.

I did not look at them as you created the impression that you are not 
interested in my code either.

> The item above was just a mistake.  Sometimes mistakes make it through
> code review.  Shit happens.  Get over it.
>
> In this case, it was found early, and I've integrated the fix for it
> already.
>
> Can we *please* move on?

Moving on the way I understand it is a bit different than Illumos currently 
works. Let me make some examples: Your work has the same unimportance as e.g.
star, a pax replacement via star or the OSS cpp I recently created... unless 
you value other people's code the same way you value your own code, you will 
have problems to get the attention of other people. Let us treat everything 
and everybody the same and things will change.

If a program (des) that exists since more than 20 years and that even has a
CDDLd man page is just missing a package description, then the right way to 
deal with this problem is to create a package description and not to remove 
the program. Could we come to such an agreement?

I am not sure whether you noticed that I am working on making OpenSolaris 
manageable as a real (selfhosting) OSS project. Everybody is free to 
collaborate with me as I am collaborating with anyone who is interested and I
am opening my code at the same time I create it.

> Btw, I will *remind* folks.  Prior to this the implementation of printf
> we had was *CLOSED SOURCE*.  So, some bustage occurs in the process of
> opening it up.  At *least* know we can look at the code.
>
> If you don't like what I'm doing, feel free to fork.  At least thanks to
> my extensive efforts you can *do* that now.  It was not all that long
> ago that forking was not a practical option.

If you do not treat everybody (including you) the same way, I may be forced to
go this way. I do not intend to do so as long as there is hope for 
collaboration. Do you still like to collaborate, so please show some hints.

I am interested in an OpenSolaris code base that is as POSIX compatible as 
possible and that is as compatible to previous Solaris releases as possible.
As you might guess from the way SchilliX works, I am also interested in an 
OpenSolaris distro that is following UNIX and Solaris traditions. I have a 
concept on how a svr4 pkgadd based distro could download everything from the 
net and I plan to make SchilliX to become such a distro.

I am a user of UNIX compatible systems since 1982. I started to write 
UNIX applications in 1982 (star and the first shell with history editor). I 
started to write kernel drivers in 1984 (helped to convert a UNIX clone to 
convert from swapping to virtual memory and I wrote the ptrace syscall support 
that is different in a paging VM system). I worked on the first Sun machine that 
came to Europe and I wrote generic SCSI drivers and the first layout 
switchable console keyboard driver for SunOS in January 1986. I invented "fbk"
the file->blockdevice emulation layer in Autumn 1988 and I wrote the first COW 
filesystem (WOFS) in 1988-1991. I wrote a segment driver for a S-BUS<->VME-BUS 
adaptor in 1990 (to connect the 256 MB of RAM from our image processing 
Hardware to Solaris) and I wrote the Joliet support for "hsfs" in Solaris. I 
modified the FORE ATM driver in 1996 to support speudo multicast connections 
for the MBONE. I have a lot of knowledge on the SunOS internals that people 
miss who started recently with SunOS. 

I could do code reviews for a lot of projects and I could on general help a lot 
if people are interested in my skills.....

You of course would need to ask yourself how you could help others and how work 
could be distributed amongst people.


Let us take this discussion as a initial ignition for a better collaboration in 
the future and let us try to avoid the need for a fork.

Jörg

-- 
 EMail:joerg at schily.isdn.cs.tu-berlin.de (home) Jörg Schilling D-13353 Berlin
       js at cs.tu-berlin.de                (uni)  
       joerg.schilling at fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/
 URL:  http://cdrecord.berlios.de/private/ ftp://ftp.berlios.de/pub/schily