Re: Mutually dependent libs double loading.

On 9/23/05, TRANS <[e-mail]> wrote:
> I'm very suprised I have not gotten an official answer about this. Is
> this really an ignorable issue? Apparently most of what I have to say
> is ignorable. Well, at least Florian is generally kind enough to
> comment. Thanks Florian. I realize people are busy --we're all busy.
> Is the problem that only Matz and Nabu have any authority? (or know
> enough english to respond?) Do I not follow proper protocols? Is that
> it? If so I wish some one would direct me to them. Or is it just
> something one's supposed to "know" being with the "in" group. Well, I
> don't do the "network" thing, so I guess I'm out of luck. or is it
> that it's just a dumb question --and most of my thought equaly as
> stupid. Maybe that are. If so I wish someone would just tell me so
> instead of letting rattle on like I'm actually doding something.
>
> Okay, that's my rant. Yes, it's a combination of unspokens building
> up. I think the camel straw was not being able to get a list of Kernel
> module_functions, and that upon discovering, after years of thinking
> otherwise, Object doesn't really have any methods.
>
> T.
>

I share some of this sentiment, though, I am also aware of why they
can't always answer: This list is high traffic. I have had a re hard
time tracking everything lately so I am starting to let a large volume
pass before I read a topic. This would be an example of that case.

To expand on your encountered behavior...

This has been posted about before, though in many forms. Basically the
require mechanism is very simple as is. Saying that ruby should add
the main script to the list of required ones is not as simple as it
may sound at first. The reason for the behavior that you are
experiencing lies behind this:

$ cat t.rb
p 1
$ ruby
require 't'
require './t.rb'
require '././t'
# These will not load the file now:
require 't.rb'
require './t'
require '././t.rb'

As you should note, ruby detects includes by the actual path given to
find it. When the '.rb' is left off it, it is still equivalent to with
it on so you will no extra require.

So this is also why the program is hard to add:

ruby ./script.rb
ruby script.rb
ruby `pwd`/script.rb

and so on...

These are all going to be different. What if you have a script with
the same name as your main script, well ruby will most likely allow it
to load BUT this is the side effect.

Suggested solutions have been posted too. I suggest a search but to
reiterate one that has come up many times:

Tracking by inode or equivalent for each platform.

I am fine with how ruby is, it is dangerous though, so step carefully.
There are a few hacks around this. These are also archived. The basic
idea is to leave a mark when executing the first time and then
detecting that mark the next time you load.

Hope this helps a little.

On the topic of how the ruby community has been evolving...
The evolution of how people tend to react has become more hostile and
less forgiving. I think this might have to do with the number of
trolls and the number of repeated questions. It is nice to see those
times when people actually try hard themselves first before posting to
the list. However, it seems the nice days are long past (not that I am
senior as some of the others here).

This does not rule that our mentality has to remain this way. It is up
to each individual to act as they please. If we want our outstanding
community to continue in excellence it is each persons own choice.

Brian.