OpenEuphoria

2. Re: IDE and eu4

• Posted by AndyDrummond Jul 12, 2009
• 1260 views

It looks to me - as I wrote xpmmer - that something has happened to the source of IDE_xpmmer since the procedure is just a normal one. You seem to have a newline before the trailing ) which would rather upset the parser, I suspect. There really should be no reason why the procedure argument list should cause a fault.

Andy

4. Re: IDE and eu4

• Posted by mattlewis (admin) Jul 12, 2009
• 1294 views

No, it's a library function. The key to this bug is that you have what is effectively the use of a forward type (int). It's declared somewhere as a global type, but that file isn't included by IDE_XPMmer.ew, which means that the interpreter cannot resolve it until the end of parsing.

The other key ingredient is that the parameter name is the same as some other routine. Here is a minimal example to display the bug:

procedure bar() 
end procedure 
 
procedure foo( int bar, int what ) 
end procedure 
 
type int( object i ) 
	return integer( i ) 
end type 

This has been entered as bug 2820451:

https://sourceforge.net/tracker/?func=detail&aid=2820451&group_id=182827&atid=902782

I seem to recall that Derek said he was going to be working with the IDE. Derek, have you done much with it? I looked at the sourceforge project page, and it doesn't show a svn repository, and the cvs repository is empty.

Matt

6. Re: IDE and eu4

• Posted by mattlewis (admin) Jul 12, 2009
• 1258 views

int is not the same as integer. There are many optimizations in the backend when it knows that objects are integers. It generally can't do this for user defined types, which int is.

As I explained, the parser wasn't dealing properly with a forward type when the parameter name was the same as a previously defined routine name. It was a bug that it gave an error in this situation. It's been fixed in r2212.

Matt

8. Re: IDE and eu4

• Posted by DerekParnell (admin) Jul 12, 2009
• 1243 views

Why do you suspect the intention of the type definition was to duplicate the C "int" on 32-bit machines?

As far as I recall, it was developed only as a shorter alias for integer; the coders got lazy

10. Re: IDE and eu4

• Posted by SDPringle Jul 12, 2009
• 1239 views
• Last edited Jul 13, 2009

With my trusty editor I was able to search and replace 'where' with 'what_place' and then 'what_place(' with 'where('. I searched and replaced 'routine' with 'a_routine.' I did this using regular expressions so that I made sure these were whole words. This was enough to allow eui -test to run through without an error. I typed the following no the command line and pressed ENTER four times:

time && eui -test IDE.exw && time 

I had a long bath. When I came back to the computer this is what I saw:

La hora actual es: 17:46:30.93 
Escriba una nueva hora: 
La hora actual es: 18:07:41.54 
Escriba una nueva hora: 

This means it took twenty minutes for eui to check the file and it never even started running IDE.exw. This is on a Pentium IV. So, I stearted it again without the test switch and I started preparing supper. After I had finished supper, I had the IDE up and running. It doesn't run slowly. When I try to run the project it says CreateProcessHi failed.

Shwan Pringle

12. Re: IDE and eu4

• Posted by jaygade Jul 12, 2009
• 1329 views
• Last edited Jul 13, 2009

procedure bar() 
end procedure 
 
procedure foo( int bar, int what ) 
end procedure 
 
type int( object i ) 
	return integer( i ) 
end type 

Wait a C int hasn't been -32,768 - +32,767 for a long time, at least since the advent of common 32-bit systems. At least 10-15 years now...

For most non-DOS systems, a C int is -2,147,483,648 to +2,147,483,647. Which includes most computer systems in current use.

Not that using legacy systems is a bad thing, mind you.

14. Re: IDE and eu4

• Posted by euphoric (admin) Jul 12, 2009
• 1235 views
• Last edited Jul 13, 2009

Will do! Is this going to be the official repository of Win32Lib IDE? I always thought EuVIDE was a separate project...

16. Re: IDE and eu4

• Posted by ChrisB (moderator) Jul 13, 2009
• 1236 views

time && eui -test IDE.exw && time 

La hora actual es: 17:46:30.93 
Escriba una nueva hora: 
La hora actual es: 18:07:41.54 
Escriba una nueva hora: 

Hi

I believe that the reason IDE is taking so long to fire up is because of the two ways (from a simplistic point of view) that eu now includes sub-files.

The quick way - each sub included file now explicitly includes the files it needs.

The slow way - each sub included file doesn't include the files it needs so for each symbol the interpreter searches the entire program tree.

(I know this is a simplistic view, but it does explain the slowness of some programs start up)

If win32lib and ide have not been updated with these explicit includes, then the startup is significantly slower. I do not think they have yet been updated (correct me if I'm wrong). Incidentally, this also affects any programs created with win32lib. I hope they will be much faster once they are. Until then I will continue to use the pre compile IDE available in the archives.

Chris

18. Re: IDE and eu4

• Posted by mattlewis (admin) Jul 13, 2009
• 1274 views

C:\win32lib-ide\includes\ide_xpmmer.ew:3214 
badly-formed list of parameters - expected ',' or ')' 
procedure drawShape(int where, int what, int x1, int y1, int x2, int y2, int Dot) 

Are you certain of this? What happens when you run the test code from bug 2820451:

procedure bar() 
end procedure 
 
procedure foo( int bar, int what ) 
end procedure 
 
type int( object i ) 
    return integer( i ) 
end type 

Matt

20. Re: IDE and eu4

• Posted by euphoric (admin) Jul 13, 2009
• 1265 views

C:\win32lib-ide\includes\ide_xpmmer.ew:3214 
badly-formed list of parameters - expected ',' or ')' 
procedure drawShape(int where, int what, int x1, int y1, int x2, int y2, int Dot) 

procedure bar() 
end procedure 
 
procedure foo( int bar, int what ) 
end procedure 
 
type int( object i ) 
    return integer( i ) 
end type 

Matt, I just tested this at work and I get the same error (here I'm using r2215).

When I run the test code above, nothing happens (I don't get an error).

22. Re: IDE and eu4

• Posted by mattlewis (admin) Jul 13, 2009
• 1208 views

r2216 should fix this now.

Matt

24. Re: IDE and eu4

• Posted by DerekParnell (admin) Jul 13, 2009
• 1223 views

I'm working on that.

26. Re: IDE and eu4

• Posted by mattlewis (admin) Jul 13, 2009
• 1205 views
• Last edited Jul 14, 2009

Like I asked a long time ago: How could that possibly work?

Matt

28. Re: IDE and eu4

• Posted by SDPringle Jul 13, 2009
• 1184 views
• Last edited Jul 14, 2009

The standard library often has forward referencing but any legacy EUPHORIA program will ofcourse not be using it and thus there would be no problem. I'd love to have 4.0 forward referencing and the regex stuff in a dll rather than linking statically. However, when an approach that performs so much better than an artistically superior solution one should compromise and give up what is more artisitically superior in favor of what has more practical value.

The question if the interpreter had a 'no-forward-referencing' switch would you be able to cut down this parse time significantly? I guess it would take a significant amount of coding just to find out. If you reduced it by 99% it would still be too slow: 20 seconds. If we don't make this fast here, 4.0 will need pre-compiled headers.

Shawn Pringle