Re: [tlug] TLUG Site with Hakyll Update

[Curt Sampson <cjs@example.com>]

On 2019-03-17 00:55 +0800 (Sun), Raymond Wan wrote:

> On 13/3/2019 8:01 AM, Curt Sampson wrote:
> > Just remember, in my world there's always a top-level `Test` script
> > in the repo, that should usually just do everything necessary...
>
> I'll keep that in mind; thanks for making it easy to start!
> Something I should consider for my repositories.

Ya, it really helps everyone. A good way way to get started on it is
just to think to yourself, every time you're about to add a line to
the README, is, "could I make this documentation executable?" and see
if you can add a line to the top-level "test/build/run/setup/whatever"
script instead.

> There was a time when Debian/Ubuntu system administrators
> were advised to favour installing Perl/Python packages over
> their CPAN/pip equivalents.

Sysadmins, yes, I can see that: I'd even advise doing so today for any
dependencies that the particular application you're using doesn't
handle for itself. But that's quite a different world from
development. Think about it this way:

- Sysadmin: "I need this to run on my host. I don't care about anybody else."
- Developer: "Nobody cares if it runs on my host."

Though these days even sysadmins are being provided and like to use
tools to do independent installations of dependencies for applications
(e.g., Docker containers). Trying to integrate the Apache configs for
six different applications is an error-prone pain the you-know-what.

Even since long before Docker, back in the early 2000s, I'd caught on
to this way of doing things because of my long background in sysadmin.
As a developer I would commit Apache and other configuration and
provide scripts for developers to start up their own server processes,
independent of others running on that machine, servers for both
automated and manual testing.

Remember: "Configuration is Code."

> > Oh, static website builders like this are still very much a
> > "Content Management System"; just one that keeps the source code
> > for the site in a Git repo rather than a MySQL database.
>
> Ah...  Yes, quite true.  I mistakenly assumed that CMS' were those
> that were built on databases, but indeed, MySQL isn't the only way
> to manage content.

Yes; this is really about stepping back and thinking of the goal,
rather than the implementation. The same problem has cropped up with
"continuous integration". A lot of developers seem to have heard about
the idea of CI only since the advent of automated CI systems (Jenkins,
etc.) in the late 2000s and so they assume that CI means "automated
independent build robot," but actually the whole idea of CI started
long before that (starting to come to broader attention in the late
90s, with Extreme Programming, I think) and I've worked on tons of
projects with multiple full integration tests done every day and yet
no automation on it.

cjs
-- 
Curt J. Sampson      <cjs@example.com>      +81 90 7737 2974

To iterate is human, to recurse divine.
    - L Peter Deutsch