Ok. You were lucky to have chosen '==' as an object name, then.
The insanity begins, when the object names use characters that are forbidden by some file systems. On Windows, for instance, you can't have ? < > | : * \ " in filenames. This means, if you want to be able to load such an object, you need some mechanism to substitute those characters by something that is allowed.
This is where hexloader comes into play. It was necessary because Pd-extended decided to split libraries into single-object-files, so that each object has its own dll|pd_darwin|pd_linux file. However, it's not possible to have >~.dll. The problem got worked-around by substituting the forbidden characters by their hexadecimal representation. I don't know the exact details, but hexloader enabled Pd to search not only for the name you entered into the object box, but also for its hex counter part. This means you could create an object [>~] and Pd would load 0x3e0x7e.dll, if it was available.
All this caused some degree of confusion, since hexloader wasn't loaded per default in Pd-extended. This means, after [import zexy] you were able to create [symbol2list], but not [>~]. You had to [import hexloader], too, to be able to use [>~]. This created troubles for quite many users. (Then there was the issue with object name aliases, so that you could load [s2l] only after having created [symbol2list], but this is another story).
All this was never an issue with the canonical zexy library, because everything is packed in zexy.dll|pd_darwin|pd_linux. Once you load zexy, all its classes simply worked.
I maybe was over-dramatizing the situation, since your example with [==] seems to work fine. However, since Pd-extended has reached its end of life, we should take the chance to learn from its mistakes and I consider the forceful splitting of libraries a mistake and the proposed solution with hexloader a kludge. It might would make things a bit more straight forward, if hexloader was part of Pd, but that's not likely to happen (I remember Miller having expressed hesitation). And the whole hexloader thing is currently not necessary as long as external authors put their classes with 'weird' names into multi-object libraries with 'sane' names or use only 'sane' names for their single-object externals.
I hope I could make things a bit more clear and my explanations stick to the facts.
Roman
On Tue, 2016-04-05 at 11:43 -0300, Alexandre Torres Porres wrote:
2016-04-05 5:08 GMT-03:00 Roman Haefeli reduzent@gmail.com: If you're simply interested in knowing how things work technically, fine.
I'd love to know, for sure, that's why I'm asking :)
Now that we have a chance to get rid of all hexloader related kludges, now you come and bring it up again.You see, I don't really get what you mean by "hexloader" or its related kludges. All I know is some [hexloader] object that is in my pd extended 0.42-5, and all I know is that I need to use it in order to load the [==~] object from zexy. What you're talking about, somehow, relates to that?
Anyway, seems so to me... and if so, the thing is that what I'm asking and doing has nothing to do with "hexloader"... (I never even mentioned about "hexloader", btw) ... and I read about the "hex loader" discussion as suggested, and found stuff that I didn't really think was related to my questions. Yeah, like I said, I don't really know much and I'd like to know, so I might be missing something, and someone can help me with it...
But the thing is, all I asked was how to compile an object like [==~] and make it load without being part of a library. I found on deken a zexy version that seemed to do that (specifically: zexy-v0-0extended-(Darwin-i386-32)(Darwin-PowerPC-32)(Darwin-x86_64-32)-externals.tar). And it didn't need a [hexloader] object too, by the way.
I didn't get an answer, but me and my colleague were checking the source code from zexy and found some cues. We tried it... and it works!
Now I have an object that is compiled as [==~], it's not part of a library, and it loads and works on pd vanilla 0.46-7 64 bits, pd vanilla 0.46-7 32 bits and also Pd-Extended 0.42-5 (without the need of the [hexloader] object by the way). All you need is the ==~.pd_darwin object in a search path.
Speaking and thinking as a user, I think it is easy and great to have a working and compiled object that just loads and works, so that is what I 'm after.
But anyway, yeah, I wanna know what are the dangers and all...
cheers