Re: [Evolution] Issue with formatting of gconf files.

[Pete Biggs <pete biggs org uk>]

>       HTML code is bad to begin with due to its fundamental formatting, but
> leaving out linefeeds, tabs, and spaces just makes it worse.  Would you
> like reading a book with no punctuation, no paragraphs, and lack of
> structure?  Yet I see lots of that kind of documentation, and even more
> code like that.  This is not furthering the profession, nor is it
> helpful.
>
> With that said, here is a snippet from the gconf stuff for evolution:
>
> Directly from the file:

DON'T PLAY WITH THE GCONF FILES.  They are NOT meant for human
consumption.  They are machine readable XML files formatted in a way
that is most convenient for a machine to read. The only reason
the .gconf hierarchy exists is as a dump of the current config between
sessions & the only reason they are text based XML is that it is an
extensible standard that is easily parsable. It matters not one jot to
the machine if there are line breaks or tabs in the XML to make it look
pretty - all it does is to take up extra space.

> How many of you can read and understand this bit of code?

My gconfd can and that's all that matters.  They are NOT text config
files like /etc/yum.conf.

> I especially dislike the ########## ##### # localhost localdomain
> filename.

It's not a file name, it's a UID - the clue is in the bit before it
which says "uid=".  It's a unique identifier for, in this case, a group
calendars.

>   This has no significance to the job being accomplished.  As
> much as I rail against "self documenting code", this is just beyond
> useless.  It smacks of trying to use obfuscation for security, which has
> been proven over and over to not aid security.  Other than that it has
> no value, no significance and adds no real value to the process being
> described.

Since these are uids are never externalised by the program, there is no
significance, or value, in giving it a human readable name.


> But putting an entire page of a book of code in one line is
> unforgivable.  I write converters between systems, changing hardware,
> software platforms and instrument capabilities.  I would be embarrassed
> to show something like this to one of my customers, moreover it would
> probably result in my never being hired again.

The fact that you seem not to grasp the usage or significance of Gconf
and are willing to indulge in a rant without researching what you are
ranting about would make me very worried if I were one of your
customers.


>       Cleaned up just a bit, the use of the code becomes clearer.  But it
> also uses two forms of HTML controls, both the <header>...</header> and
> the <header />.  Both are of course legal and useful in some contexts,
> but it is usually preferable to maintain consistency.

IT'S NOT HTML.  Not everything that has <..> in it is HTML.  And both
those contructs you mention are used consistently AND properly in the
XML snippet you are quoting.

> If you read all the way to here, thank you.  If you disagree, that is
> OK, but remember that one day you will have to support code written in
> one block that is over 3000 characters long (I have already seen that on
> some web pages.)

It's NOT code either - it's XML.  It's a data description language.

P.