[KLUG Advocacy] What users want from Linux?

[Adam Williams]

> >Here is a supposedly positive (overall) article on Linux adoption.  It
> >starts out with saying that the biggest hurdle is probably FUD -
> >http://www.nwfusion.com/news/2003/1020linux.html?docid=8139
> >- then goes on to reveal itself as so hopelessly unresearched that it
> >itself might qualify as FUD.  
> Clueless cheerleaders?



> >I just want to know where they find these experts?
> Actually, they make themselves available. When experts don't have full-
> time work experting, they have to write articles to keep the wolves 
> from the door, so we get these things. Sometimes those who can't do, 
> write.

I've been called several times by trade rags and asked for quotes;  as
far as I know they've only been used once - and they got what I said
wrong.  I had the impression that they're were people who trawled for
statements, etc... from which to assemble articles.  Many articles read
like they were assembled in such a manner.  I didn't they these things
got "written" in the old sense of the word;  articles in
InfoWorld/NetworkWeek seem different in form than submitted artciles
like one finds in LJ.  My beef was they just seem to use the collected
material without doing much if anything to verify or cross-reference it.

> >Half the people at a KLUG meeting are more knowledgeable than
> >these twerps.
> Maybe we would do well to get them to write articles!

Not a bad idea,  except I'd have no idea how to go abou doing that.

> Overall, Adam has provided great comments. I'll mostly add to them, and in
> general, I'll point out that general perceptions lag reality by AT LEAST 
> 18 months, and twice that isn't unusual.
> In the interest of brevity, I'm deleting anything I'm not commenting on....
> >1. Lack of monitoring tools.
> >You've got to be kidding!  From OpenNMS to Trivoli, everything supports
> >Linux if Linux isn't the tools native platform.
> Lack of research, maybe they don't know where to start (glossy sales
> stuff from IBM, maybe freshmeat.net?)

Could be, but how?  I get stacks of promotional mail from companies
(including IBM, Steelhead, etc...) that would be happy to provide a
monitoring solution for my Linux systems.  It even comes to my house.

> >2. Reliability
> >I like the quote "Commonly used applications such as OpenLDAP need to be
> >made cluster-aware.". &*&^(@ *$&*(@ $(@*$)@@#$_$%@ #$*@*#$&(  Can't
> >these "experts" even bother to subscribe to the Linux Journal which has
> >covered High Availably LDAP at least twice.
> They read something about openLDAP two years ago, didn't bother to 
> come back (might be a good percentage play, given the pace of change in
> commercial software of comparable size and complexity). The pace of OSS
> is different, they're not used to that yet...
> It's easier to write than it is to do research... 

Apparently.

> those darn search engines are so user-unfriendly, and who needs 'em when the author 
> "just knows" the truth?

The Oracle of Borland Delphi?

> >6. Cohesiveness
> >Whatever this even means.
> Is this the same as "Integration"?

Beats me.  This is one of these words that if not followed by at least
three paragraphs of explanation doesn't mean anything.

> >7. Applications
> >"Support for Windows-based applications would be a definite plus," God,
> >No!
> This is one of those ideas that seems simple and obvious to the user.
> The devil is in the details of course, and it's another symptom that 
> people simply don't want to see any alternatives.

Thats my opinion.  Bringing Win32 apps to Linux would just bring Win32s
problems (AND COST) to Linux.  If you want to run a Win32 app run the
Win32 OS.  It is a failure to see applications as solutions to problems;
instead seeing one application as "the solution".

> One thing people who make this claim do not understand is that there are
> a LOT of ideas and behaviors that are not and never have been part of 
> the Windows API and environment, and only a subset of that has made it 
> into recommendations, coding standards, and style guides. Making this
> claim effectively shuts out all of that.

And ultimately the list of niggling little things that need to be
'emulated' is almost infinite;  making it impossible to provide a
"cohesive" :) user experience.

>  I will leave further commentary
> on the behavior and performance of "Windows-based software" as an
> exercise for other participants in this thread, or for the reader.
> Frrankly, there are a lot of things about X-based desktops that I prefer
> over the way Windows works, because they save time and allow me to be more 
> productive. There's room for imporovement, but for me, the race has been
> won.

Same.

> >8. Skilled Developers
> >"Many universities just teach students how to program on Windows, and as
> >a result they don't get to learn much (if anything) about how a computer
> >really works.
> Hogwash! Places that "teach people how to program" are called "Trade
> Schools". I don't care what they call the credentials given out on
> completing the coursework. There is a deep misunderstanding of what
> higher education is supposed to do for people...

Maybe the issue is what is "higher education".  Maybe a community
college just isn't.  Thats the conclusion I'm coming to.  I doubt places
like GVSU/Calvin/etc... are either.  Glorified trade schools. (While
GRCC is a glorified High School).

> >Then when it comes to debugging race conditions, network programming, or 
> >library code they don't have a clue about what to do."
> Good university software engineering and CS curricula teach principles,
> recommend programming environments, and expect students to pick up the
> many of the details. I'd like to see CSE folks on this list write some-
> thing more about this.

Must be true, as some decent software gets written.  But it would be a
very-good-thing if some of these practives would get shoved down the
food chain a bit;  as not every project justifies the $$,$$$ of a true
CS body,  but they could certainly benefit from better practices.

> >Yes! Yes! Yes!  But what does this have to do with Linux?  They can't
> >program worth a crap on Windows either?  
> My general sense is that a lot of low-level programming instruction has
> been deprecated... relegated to trade schools and community colleges,  or
> really special courses where thingslike assember and fairly metallic C
> is required to handle the content, for whatever reason.

I think you're correct.

> On the whole, competent programming is a skill-set that is to a great 
> degree independent of platform. That said, good craftsman prefer better
> tools... and the incompetent generally could care less.

So true,  this statement should be on a plaque somewhere.  

Maybe in Latin so the incompetent can't read it and aren't offended. :)

> >How to you graduate from a University with a developer focused program 
> >AND NOT KNOW HOW TO USE VERSION CONTROL SYSTEMS????
> Well, there are not a whole lot of "developer focused programs", instead,
> you have a rather large number of places that teach these things bit-by-
> bit.. one class on understanding and creating algorithms, another on 
> optimization, others on I/O systems and economics, others on human inter-
> face design, and so forth. 
> Again, this is a good place to read some comments from CSE practitioners.

I suspect current programs are much to student-friendly for this too
happen.  Colleges (I've been told by administrators) are all fighting to
be the most student-friendly in regards to scheduling and course loads, 
so the general effect is more fragmented programs and a general
dumbing-down (making pre-reqs softer, which kills the effectives of
higher level classes).

> One approach I've observed is that some programs are doing more role-
> play in senior classes, where the goal is NOT "study the characteristics 
> of these algorithms" but rather "Deliver the application". This leads to
> a lot more holistic experience, putting together the diverse skills needed
> in the real world, like writing specifications, communicating with users,
> and going through collaborative development, which tools to match. Perhaps
> this practical approach as the final layer of a good education in soft-
> ware engineering will deliver people into the industry with a better 
> understanding of what they must do to be productive.

That sounds like a good plan.  I had one instructor who did that,  but
since the project was limited to the single course the scope was rather
narrow.

> >9. GUI
> >http://www.ximian.com
> 18-36 month lags, maybe more. The drumbeat of "Linux is CLI-only, like DOS"
> is silly and harmful, regardless of the merits of a particular desktop.

"Silly" and TIRED.  Even the same magazines that print this crap have
run articles about the advancements on the Linux desktop.

> >10. Consolidation
> >http://www.ximian.com
> >Who can't come up with ten problems that plague Linux?  But this is a
> >list made by people who don't know what they're talking about,  and it
> >gets printed, and they get paid.  What a crock.
> It does keep them from doing more serious harm, although this stuff
> seems bad enough....

More annoying than I think actually harmful; a real waste of resources
and page space.  At this point I suspect there are people open to Linux
(likely to dismiss this as tripe they've read 100 times before) and
people who think M$ should just whack Torvalds and get it over with (who
won't read an article about Linux anyway).  Thats the two camps I
encounter,  rarely does one find a "Linux, huh?" person anymore.