[OT] Grumble...

[James Kanze]

I must have somehow got into people's killfile without ever having
posted so much in the NG, but I will persist. James, this "scripting
tasks" that you keep talking about, have you ever considered writing
them with Powershell?

It's not installed on my windows machines.

Seriously, it could possibly be an alternative, although much of
what I use CygWin for is awk or sed---the tools, and not just
the shell. Another alternative (the one that is officially
accepted here) is perl. Anytime the scripts become part of the
official build process, I have to convert them to perl.

The fact is, however, that I know AWK, sed and company very
well, and in so far as I have to install an external tool to do
the job, I might as well install one which I know. (If the tool
were bundled, or even officially installed by the company, the
issues would be different.)

And the real issue isn't CygWin vs. some other set of command
line utilities (I'll admit that CygWin has a lot of problems),
it's using command line utilities and scripts vs. some sort of
pre-digested click and point that doesn't do half of what you
need, and gets its dependencies screwed up as well as soon as
there is generated code.

MSBuild is Microsoft's version of Ant and you can invoke
Powershell scripts from that and perform whatever the heck you
want.

So I've heard. The choice of ant was at least partially based
on portability. And wasn't mine---I know GNU make, which while
ugly as sin, does get the job done. But in this case,
portability was one of the requirements---we currently maintain
two versions of the build procedure, one in VS project files,
and one in GNU makefiles, and keeping them synchronized is a bit
of a hassle. The nice thing about ant is that the input format
is simple enough that even the Windows programmers accept it.

Powershell can also be used to write repetitive tasks the
Ian Collins asked about in another thread. I noticed another poster
named Ralph also mentioned this elsethread but expectedly that wasn't
taken up either. Is there simply a refusal to look at what is
available on Windows and simply parrot Cygwin as a solution to any
problem?

No. I haven't been following this thread in detail, and only
respond to it when I happen to notice some particularly stupid
comment (and not always then, either). I know that Microsoft is
evolving, and that they do have a number of interesting tools,
but again, they're not bundled, and they're not installed here.
(The only thing that is officially installed is Visual Studios
and perl.) And since we do have to support Linux platforms, and
a few old Solaris machines as well, portability is an issue for
anything that we do have to install.

 

[Keith H Duggar]

And the real issue isn't CygWin vs. some other set of command
line utilities (I'll admit that CygWin has a lot of problems),
it's using command line utilities and scripts vs. some sort of
pre-digested click and point that doesn't do half of what you
need, and gets its dependencies screwed up as well as soon as
there is generated code.

Exactly. And it applies across a broad spectrum of computational
tasks. For example, back in my computational biologist days, I
taught an open minded friend the paradigm and he embraced it.

Today his bioinformatics colleagues are simply floored again
and again by how flexibly, how massively, and how quickly he can
perform tasks with some perl/awk/sed/unix Command-Line-Kung-Fu
compared to their Excel/point/click/windows Gui-Rsi-Slogs.

That said, it's simply hard to appreciate the differences and
truly understand where each (command line or gui) is superior
unless you have sufficient experience with both. Nearly everyone
has sufficient experience with GUIs. Far fewer with Command-Line-
Kung-Fu.

Yes, like Kung-Fu, Command-Line-Kung-Fu takes years to master.
Yet ultimately you become a hybridized cybernetic warmachine.

I say "hybridized" because the optimal solution is to use BOTH
CC and GUI. (But, the sheep majority are skewed to GUI-fanboys
with zero critical thinking on the "choice".)

KHD

 

[Gennaro Prota]

On 17/08/2010 11.55, James Kanze wrote:
[...]

So I've heard. The choice of ant was at least partially based
on portability. And wasn't mine---I know GNU make, which while
ugly as sin, does get the job done. But in this case,
portability was one of the requirements---we currently maintain
two versions of the build procedure, one in VS project files,
and one in GNU makefiles, and keeping them synchronized is a bit
of a hassle. The nice thing about ant is that the input format
is simple enough that even the Windows programmers accept it.


No. I haven't been following this thread in detail, and only
respond to it when I happen to notice some particularly stupid
comment (and not always then, either).

*Ditto*. The else-thread mentioned above was in reply to me. And
seemed just in the spirit of pointing out an ulterior tool ("you
might take a look at Windows PowerShell") so I didn't feel like
replying with why *I* do not use it.

But since Dilip is suspicious ;-), here we go. As far as I'm
concerned the issue is:

they tend to invent the world and by the time you have learned
something on it they already have a new beast. And I aim at
portability. So, while I might take a look at their proprietary
things, they are almost never the solution of choice for my
programming tasks.

Satisfied, Dilip?

 

[joe]

jacob navia = troll. That summarizes all my replies

"Ya'll" say that anytime "ya don't get" something. "Trolls" don't
modify/repurpose compilers, ya think?