Misc questions

Misc questions

Post by slavin.. » Sun, 31 Dec 1899 09:00:00



Hello,

I've got 2 questions:

I have an APC SmartUPS and a modem on two different serial ports (and
different IRQs), yet I can't monitor the UPS and dial out at the same
time.  The serial support is compiled into the kernel.  Are there any
special serial options that can be specified before compilation to
alleviate this problem?  I looked through serial howto, but didn't find
anything appropriate to my situation.

Another question:  I've been looking through Linux source, and needless
to say, the code lacks documentation somewhat.  Some files are better
documented than others, but overall, IMO, developers are not commenting
their code in enough depth.  Isn't it one of the prerequisites of good
software development?

Sent via Deja.com http://www.deja.com/
Before you buy.

 
 
 

Misc questions

Post by John Hasle » Sun, 31 Dec 1899 09:00:00


Quote:> I've been looking through Linux source, and needless to say, the code
> lacks documentation somewhat.  Some files are better documented than
> others, but overall, IMO, developers are not commenting their code in
> enough depth.  Isn't it one of the prerequisites of good software
> development?

How many patches have you submitted so far with comments that are, IYO, in
enough depth?
--
John Hasler

Dancing Horse Hill
Elmwood, WI

 
 
 

Misc questions

Post by slavin.. » Sun, 31 Dec 1899 09:00:00


Quote:

> How many patches have you submitted so far with comments that are,
> IYO, in enough depth?

And why do you think I should submit patches before being eligible to
read well-commented code?  Please understand, I'm not flaming developers
or contributors, but I just thought the code would be easier to
understand if it were commented a little better.

Sent via Deja.com http://www.deja.com/
Before you buy.

 
 
 

Misc questions

Post by John Hasle » Sun, 31 Dec 1899 09:00:00


Quote:slavinger writes:
> And why do you think I should submit patches before being eligible to
> read well-commented code?

Possibly because I assumed that only someone who felt qualified to read and
write well-commented code would feel qualified to comment on the quality of
the commenting.

Study up on operating systems and some of the stuff that now seems obscure
will become self-evident.  You can then send in patches documenting the
rest.

For myself, I know just enough about the subject to be dangerous.  I'm no
kernel hacker.
--
John Hasler

Dancing Horse Hill
Elmwood, WI

 
 
 

Misc questions

Post by slavin.. » Sun, 31 Dec 1899 09:00:00


Quote:> Possibly because I assumed that only someone who felt qualified to
> read and write well-commented code would feel qualified to comment on
> the quality of the commenting.

Anyone who is qualified to read code is entitled to read well-commented
code.  I generally over-comment mine, but then I don't want people to
wonder about what this loop does.  It may seem obvious to me, but not to
the person maintaining it.  Especially when talking about
architecture-specific things where things are expected to be done a
certain way, and expectations vary by vendor.

Quote:> Study up on operating systems and some of the stuff that now seems
> obscure will become self-evident.  You can then send in patches
> documenting the rest.

My primary interest lies in boot loaders.  Linus's code is one of the
better-documented, but some places seem to lack a little.  Recently I
had to look some things up in Matrox frame buffer source.  Being new to
this area, it takes me much longer to decode without comments than with
them.  And there's always "I _think_ they're trying to do this here",
instead of /* we're doing this here */.  See where I'm going with this?

Sent via Deja.com http://www.deja.com/
Before you buy.

 
 
 

Misc questions

Post by Peter T. Breu » Sun, 31 Dec 1899 09:00:00


: > Possibly because I assumed that only someone who felt qualified to
: > read and write well-commented code would feel qualified to comment on
: > the quality of the commenting.

: Anyone who is qualified to read code is entitled to read well-commented
: code.  I generally over-comment mine, but then I don't want people to

  "it was hard to write, therefore it should be hard to read".

: wonder about what this loop does.  It may seem obvious to me, but not to
: the person maintaining it.  Especially when talking about
: architecture-specific things where things are expected to be done a
: certain way, and expectations vary by vendor.

Once you get the hang of the kernel architecture, the code is pretty self
evident. If there are comments, all the better, but it's usually clear to
anyone who has the capacity to program in C in the kernel just what a
loop does! What comments can help with are to say what is NOT in the code,
such as "we have to avoid freeing the master bh because there's a kernel
thread that later walks the bh list and *s any old bh's that it finds,
and it'd get sick if we killed it here".

: them.  And there's always "I _think_ they're trying to do this here",
: instead of /* we're doing this here */.  See where I'm going with this?

Linus' code is indeed very good. Yes, there is stuff I can't understand,
particularly assembler. Don't mess with stuff you can't understand. The
commenst won't help you. If you figure it out enough to comment it, you
probably want to send in a code patch making the code clearer, rather than a
comment patch making the comments on the obscure code clearer.

Peter

 
 
 

Misc questions

Post by Lew Pitche » Sun, 31 Dec 1899 09:00:00



> > Possibly because I assumed that only someone who felt qualified to
> > read and write well-commented code would feel qualified to comment on
> > the quality of the commenting.

> Anyone who is qualified to read code is entitled to read well-commented
> code.  I generally over-comment mine, but then I don't want people to
> wonder about what this loop does.  It may seem obvious to me, but not to
> the person maintaining it.  Especially when talking about
> architecture-specific things where things are expected to be done a
> certain way, and expectations vary by vendor.

Anyone who has had to maintain someone elses code soon learns to
appreciate
well commented code, and tends to well comment their own code. It's much
easier to understand the goal of a piece of code (rather than what it
_actually_ does) when there are appropriate comments. _I_ tend to
include
comments that summarize the state of things at critical points, as well
as the usual "this line does this" type of comment. I find that if I
do something like...
        /*
        ** at this point,
        **      hi_num contains the index of the last element of the array,
        **      lo_num contains the index of the first element of the array,
        **      search_key contains the key to be found in the array.
        ** Now, we want to binary search between hi_num and lo_num inclusive
        ** to locate search_key in num_table[]
        */
then I can better remember my programming goal, and the next time I  (or
someone else) visit this code, I know exactly what it was that I was
trying
to do.

Quote:> > Study up on operating systems and some of the stuff that now seems
> > obscure will become self-evident.  You can then send in patches
> > documenting the rest.

> My primary interest lies in boot loaders.  Linus's code is one of the
> better-documented, but some places seem to lack a little.  Recently I
> had to look some things up in Matrox frame buffer source.  Being new to
> this area, it takes me much longer to decode without comments than with
> them.  And there's always "I _think_ they're trying to do this here",
> instead of /* we're doing this here */.  See where I'm going with this?

> Sent via Deja.com http://www.deja.com/
> Before you buy.

--
Lew Pitcher

Master Codewright and JOAT-in-training

 
 
 

1. Misc questions on a Bully-free computer

I'm contemplating new hardware and have some questions that
I don't remember having seen answered in my many hours reading
newsgroups like this and web pages like Tom's Hardware.  I
admit that I haven't reasearched the Graphics questions, so
feel free to ignore those if you want.  I'm one of those Bully
Boycotters and have some uncommon considerations as a result
of wanting an uncommon CPU, motherboard chipset, and OS.

1) Do some "jumperless" motherboards require the use of MS-Windows
or MS-DOS to set up the BIOS?  Which ones?

2) The only ATX motherboards I've seen with non-Intel chipsets
have only two ISA slots.  Can I get sound cards and extra-I/O
cards with PCI interfaces?  Are they much more expensive than ISA?

3) Do ATX power supplies blow air into or out of the case?  Any
way to reverse it?  I want all my fans blowing air in thru filters,
even if it cools less efficiently (before things get covered with
dust).

4) Any problems running Linux with the latest SIS or VIA chipsets?
A couple years ago I read rumours that Linux ran slow with SIS.

5) Since Linux doesn't use the BIOS after boot, how is it possible
for Linux to work at all with non-Intel chipsets?  Do these chips
all emulate Intel chipset control regesters, etc?

6) Does Linux support parity or ECC memory?  (Memory is so cheap
now, I'm suprised this isn't more popular.)  If so, what happens
when Linux gets an error indication?  How is the user informed?
Are there popular protocals for handling memory errors such as
saving backup and working files before shutting programs down,
etc.?  Is there any means to report which memory is bad?  Is
there any means to work around bad memory like bad disk sectors?

7) Can integrated printer-scanners be used just as well as separate
printers and scanners under Linux?  When using any of these 3 types
of devices, will many features be missing (without writing my own
programs) which the supplied MS-Windows SW makes available?

------- Bully Boycott -----
8) Do the latest AMD CPUs still have the MS-Windows logo on them?
(In that case, Cyrix will do just fine for my needs.)

9) Just curious.  Do board makers or BIOS makers have to pay
Intel/MS licencing fees to implement PCI, PnP, or a BIOS?

10) When companies advertise using "Intel Inside" or "Designed for
Windows 95/NT" logos, do they pay Intel/MS or does Intel/MS pay
them or is it a "wash"?

------- Graphics -----
11) If I just want to run 2D X-Windows at 1280x1024x24bits, is there
any reason to have more than 4 Meg video memory?

12) Can I resonably expect Linux and/or SW running on Linux to take
advantage of AGP hardware in the next two years?

13) What's the best reasonably priced Linux-friendly video card for
running 2D X-Windows at 1280x1024x24bits (4 meg)?
How about 1600x1200x24bits (8 meg)?

14) Is there any reason to get a typical 3D video card for a pure Linux
machine?  I understand that the only 3D software on Linux for now
requires the rather rare OpenGL compatible hardware.  Is this correct?

Please don't try sending mail.  Thanks.

2. partitioning HD without destroying resident info

3. misc questions

4. Comprehensive Intro Guide to Apache wanted

5. Misc questions

6. given 5-year old Dell Poweredge 2300 server, SCO Unix, 2x9GB hDD - should I learn SCO or reformat and use Linux?

7. misc questions(gcc2.11, filenames, this group)

8. SCSI ADAPTEC AIC-7850/7870

9. misc. questions

10. Misc Questions(Bootup warnings,Colors...).

11. x setup & misc. questions

12. Pls help: Misc questions

13. Misc. questions : tty, md5sum, make