OpenEuphoria

2. Re: phix updates

• Posted by petelomax Apr 02, 2019
• 1468 views

Ah, I have a fix for that (trace on 64 bit). Change builtins/VM/Ptrace.e line 1387:

--          mov rdx,[rbx+rax*4+16]      -- rtn 
            mov rdx,[rbx+rsi*4+16]      -- rtn 

Then just recompile, p -c p, with a format PE64 somewhere in the mix.

Two long-standing problems are still preventing a release:
A memory allocation issue on linux, which means it always crashes on some systems.
A "NEWGSCAN" option that fixes a number of long-standing issues is almost in place but still has a few nasty surprises left.

The only thing I seem to be able to do with those is bash my head against the problem for a few hours, get nowhere, then leave it alone for a few days, rinse, cycle, repeat.

I have several other goodies (sorry, no list yet) lined up for the release, but there is not much point to shipping a guaranteed-to-be-buggier-than-the-previous release, so I am getting pretty frustated they cannot be released, just as much as everyone else is.

A release may also be slightly delayed by actually using curl and zip in an improved distribution mechanism, but it should be a tiny fraction of the time I have already spent on the main two issues mentioned above.

Nice to know that someone else wants a new release, anyway.

4. Re: phix updates

• Posted by begin Apr 02, 2019
• 1473 views

in pTrace.e that i have, there is on this line:

--printf(1,"active_routine = %d\n",active_routine) 
    fileno = symtab[active_routine][S_FPno] 
--?9/0 --DEV get this from prev_ebp... 
--  s8 = symtab[T_callstk]  -- {ep1,ep2,era,etd,ern,ebp,vsb_root,dcount} (see pdiag.e for full details) 
                            --   0   0   0                         0     (zero because of the si=1, btw) [DEV?] 

the line you mention is on line 1376.

is that correct to change that line?

6. Re: phix updates

• Posted by begin Apr 03, 2019
• 1393 views

hi pete,

running amb2.exw from rosetta displays:

The original: 
success: {"that","thing","grows","slowly"} 
 
Small Pythagorean triples problem: 
success: {3,4,5} 
success: {4,3,5} 
success: {6,8,10} 
success: {8,6,10} 
 
Some strange not 8 problem: 
success: {2,4} 

now i add the trace command at the very top of the file:

with trace  
trace(1)  

the result is:

pDiagN.e line 3322: symtab[220] bad S_NTyp[1] 
"sleep(5)..." 
pDiagN.e line 3064: oops, rtn[=8848] out of range[1..1285] 
pDiagN.e line 3064: oops, rtn[=9140] out of range[1..1285] 
H:\Phix\demo\rosetta\Amb2.exw:-1 (era=#00739AA9, from_addr=#0, ret_addr=#0) 
type check failure, ???(varno=8745[or_ecx=#00002229]) is ??? 
 
Global & Local Variables 
 
--> see H:\Phix\demo\rosetta\ex.err 
Press Enter... 

That is not wanted - or?

btw, may be you can release version and tell everybody about the still persistent linux problems.

richard

8. Re: phix updates

• Posted by petelomax Apr 06, 2019
• 1259 views

OK, you are right. After sleeping on it I realised I am being quite daft here.

It is not really any buggier than 0.7.9, rather I am just more aware of some issues, and I need to drop the "lone wolf" mentality on them.

I should release an "unstable" release, possibly windows-only. I'll add the known issues to the bitbucket repository, and matching threads on this forum.

I've set things going by pushing everything to the rep and started writing some release notes.