2016-02-15 18:23 GMT-02:00 Fred Jan Kraan <fjkraan@xs4all.nl>
Most of cyclone are single objects files, as per pd-extended. The nettles part is a multi object library file. This seemed the simplest way to re-introduce them while avoiding the weird symbol issues. [declare] seemed more elegant than adding hexloader compliant object file names.

I'm sorry this is  still hard for me to get, so I need to keep asking questions about what you mean. I guess I'm having a harder time understanding what you mean by "adding hexloader compliant object file names".

But I know this: I downlaoded zexy form deken and included it in the search path of pd vanilla 0.46-7. I can now easily load an object like [>~] without the need of the [hexloader] object, [declare] or anything.

There were also other given reasons, like this not being possible in a particular OS, but that doesn't seem to be a problem.

So, simply put, it is possible... why couldn't we do this with cyclone? Or is it quite possible but there is a choice not to do it?

Using declare is making it harder, so if there's a way to prevent it, I strongly believe, as a Pd user that we should prevent it. In fact, I have to confess that in 10 years of patching with Pd I had never needed to use [declare], this is the first time I had to read its help file. I have also written over 320 examples for my classes in Pd Extended and never had to use it.

I find it really problematic that it seems you have to include flags, know a weird name not related to "cyclone" and, worst of all, save and close your patch to reopen and then be able to get the object you want. This is really a bad choice over just being able to load the object without any of this.

please let me know of any issues I'm still not getting

cheers

2016-02-15 20:21 GMT-02:00 Alexandre Torres Porres <porres@gmail.com>:
On 2016-02-15 07:38 PM, Martin Peach wrote:
If you look in Pd source file x_arithmetic.c you will find this line: binop2_gt_class = class_new(gensym(">"), (t_newmethod)binop2_gt_new, 0,

Well, by looking at the nettles.c code, we find the exact same structure

sigeq_class = class_new(gensym("==~"),(t_newmethod)sigeq_new, 0,

So cyclone is taking care of this problem in the same way!

None of those names are legal file names on those systems.
 
Cool, so it's not a matter of one particular OS having an issue an not another, it's a general issue to all of them that have the workaround which is already done.

Cyclone, when it's properly set up, will also register a bunch of symbols when it starts, thereby avoiding file searches for illegal filenames.

Yep, this is what I'm assuming, that you can make it work if you do the right thing. For example, I mentioned I downloaded a version of the zexy library that loads [>~] without the need of a [hexloader] object. I just put "zexy" in the search patch and [>~] loads quite easily.
 
If only part of cyclone is loaded, it may not have registered the weird symbols, so they won't load
 
These objects are not loaded, so only a part of cyclone is in fact loaded, I'm not sure why yet, it's kinda over my head. I'm having a hard time trying to figure it out the issues of needing to use [declare] and not loading this files beforehand.

But I just believe you can have them load with no problem in any OS now, right?

cheers