Mailing List Archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tlug] Issuses posed by code reuse



On Fri, 18 May 2012 15:46:45 +0900
"Stephen J. Turnbull" <stephen@example.com> wrote:

> Nguyễn Vũ Hưng writes:
> 
>  > I love [The Mythical Man-Month] but unfortunately, I consider it a
>  > good extreme case and does not have much practical use in today's
>  > software engineering (at least for my work).
> 
> I'd have to disagree, at least for my work in open source.  The
> projects I work with whose products are damn near bullet-proof (Python
> and Mailman) have managed workflows with more or less formal
> processes.  The projects that I work with that are fragile and a PITA
> to reengage with if I just take a 6-month vacation (Emacsen ;-) have
> fly-by-the-seat-of-the-pants workflows, with everybody thinking they
> are an architect.

Well.. The problem with OSS and workflows are that they are a PITA for
anyone who just wants-that-new-and-cool-feature-in. People just want
to write a patch, send it to the developers mailinglist and get it
accepted imediatly without any questions asked. Any form of workflow
or approval process is a major setback. Yes, i know, that chaos reigns
where you have no such process, and i've seen my fair share of such
projects. But then how do you get those drive by contributors to actually
contribute without redoing all those patches yourself?


> It's not uncommon that if you're the *only* engineer on a product, or
> it's you and a couple others who are always working together, and you
> have good instincts for which boring scutwork is essential and which
> can be dispensed with because you're a tiny "solution shop", you can
> do without a defined workflow.  But anything more extensive, you're
> storing up trouble for the future in my experience.

Unfortunately, most SMBs (and big companies too!) do not have any processes
for managing code. It's written, it works, it's ok, next project.
Having someone that forms all those different pieces of code into a framework
that could be reused is ... rare... at best. Having actually a docu how
that framework works... well, i have yet to see such a docu outside OSS.

I try to educate people where i can, but it's often quite pointless.
I still struggle to get one of my coworkers to see that code review,
even if it's just a cursory look over the code, has its merits and should
be done with every project, no matter whether the customer requires it
(due to legal requirements, like aerospace or medical stuff) or not.
He just simply disagrees and sees it as a waste of time.


			Attila Kinali

-- 
The trouble with you, Shev, is you don't say anything until you've saved
up a whole truckload of damned heavy brick arguments and then you dump
them all out and never look at the bleeding body mangled beneath the heap
		-- Tirin, The Dispossessed, U. Le Guin


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links