Re: home.shtml: Make checks whether the HTML is decently generated.

[Yavor Doganov]

On Thu, Feb 07, 2008 at 06:58:41PM +0200, Kaloian Doganov wrote:
> Currently, we don't have bogus languages for articles, are we?

If you promise to remember to introduce a conditional when we add a 
bogus article, that's not necessary.  (I'll delete it, OK.)

> The whole purpose of those "test cases" may be defeated if we tend to
> introduce special processing code to handle them.  

You know, I didn't take this approach beacause of my love affair with 
the findstring function in negated conditionals.  I made this proposal 
precisely because it is less clutter and less work, and I spent some 
time seriously thinking about this (and the issues you raise in this 
message).

If you want to omit the home conditional, you have to create and commit 
the pl and zh-cn header and footer, and also their banner.xx.po and 
footer-text.xx.po counterparts in order the build to generate a full set 
of templates.  I prefer to clutter the makefile with one line than 
cluttering the repository with 12 files, 4 of which are not synced.  
Also, I find it a bit useless to validate the dummy home.pl.shtml and 
home.zh-cn.shtml -- they exist to test the rules.

> The normal flow of execution should not branch specially for those 
> "test cases" since the whole point of having test cases is to simulate 
> real cases as closely as possible.

I agree in principle, but not in corner cases like this one.

> Then we just have to keep a list of bogus files that do not contain 
> real data and should not be merged when we deploy GNUN.

I am going to do to the merge, so don't worry about this.  I agree with 
you that there should not be differences in the code between the two 
instances in trans-coord and www, but for two (or four) short and clear 
lines I'm willing to make an exception.

OK to commit only with one conditional in the home rules?