[open-cobol-list] Splitting files by platform, cobc performance survey

[Patrick]

Hi Everyone, Hi Joe

I wish I could answer your post from today but I just don't know 
anything about the CE source.

I wanted to mention that Brian Tiffin will likely have valuable feedback 
but he may be be tied up at the moment, we exchanged emails briefly a 
few days ago. Please wait for his answer, it may be some time.

I didn't want to high-jack Joe's thread so I created another one, even 
though it is similar, in some respects.

1)


I find the libcob source a bit hard to read. There is a lot of platform 
specific code within single files, adjusted with lots of compiler 
directives.

Would it be a good idea to split them off into other files, named by 
platform,  to increase readability ? I would be happy to try and help 
with this, if it was deemed to be a good idea.


2)

autotools is surely a powerful thing but I have yet to read about anyone 
that "enjoys" autotools. In fact, what seems to be repeated over and 
over again is autotools is hell. Is there another build system that 
might be more suitable? I have been trying to read about developing with 
autotools so that I can start contributing little pieces of code to the 
project. I understand the basics but there seems to be so many pitfalls 
with it and I am still afraid of falling into a hole. We really need 
more developers, if there was simpler tools that might help lower the 
barrier to entry?

3)

All of my Cobol programs are still small. Has anyone compiled a large 
codebase of let's say 50-100K lines of code? I am wondering how cobc 
performs with larger projects. If speed just isn't an issue, maybe we 
could have discussions about how to make cobc easier to get involved 
with, perhaps there are other languages that could also be used such as 
Cobol itself. There is no runtime penalty for a slower compiler.

4)

I don't really know so much about libdb, I am trying to read up on it. 
Does it have a very different purpose then let's say sqlite? It seems 
smaller and I assume faster but with Oracle changing the license and 
affecting the way we can, then in turn license our code using any 
current version of libdb, it seems like a discussion on alternatives 
might be a good thing.

-Patrick