OpenEuphoria

1. sh: 1: /media/mycomputer/Seagate: not found

• Posted by petelomax Jan 30, 2021
• 1507 views

Forked from Re: [phix] what is a hybrid interpreter compiler?

p hello.exw 

$ p -c hello.exw 
sh: 1: /media/mycomputer/Seagate: not found 

Not sure I can help much
Can you manually run the hello executable on the USB drive after the not found message?
On Windows I'd suggest ProcMon, but don't even know what the equivalent is called on Linux, though I did once run it, maybe 3 or 4 years ago.
You might be better off raising this on StackOverflow

3. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by irv Jan 30, 2021
• 1475 views

I think we would need to know the exact setup.

Just as a start, I copied some Phix/Gtk programs to an encrypted external ssd. They run fine. I also compiled several pre-compiled programs to that drive, and they also ran fine. I then compiled a few directly on the ssd, and they also worked properly.

p and the builtins are in usr/local/bin (this may be the difference)

If yours are on the external drive, as well as the programs themselves, that may make a difference. Tell us where those files are, and I can try to reproduce the setup.

5. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by irv Jan 30, 2021
• 1530 views

$ p -c hello.exw  
sh: 1: /media/mycomputer/Seagate: not found  
 
It compiles but does not also run.  

• In interpreted mode, p "compiles" to memory, and runs it from there, no problem.

• In -c mode, it "compiles", saves to disk, writes a shell script, and then calls bash to run it, one could guess, based on the sh in the error line.

Is the newly-compiled program executable actually saved on the Seagate? If so, can you run it from a terminal? If that works, then....

• The message above makes no mention of a program name, so perhaps bash first tries to cd to /media/mycomputer/Seagate and fails.

• Alternately, perhaps the script line is supposed to contain the path plus the filename, but the filename is being left off by Phix.

• Or, bash doesn't have the priviledges to run programs from an external drive (but don't ask me how that can happen)

To find out, perhaps you could write your own bash script to change to the Seagate and run the compiled program. If that doesn't work either, it's probably permissions.

I would try running sudo p -c hello.exw and see if that makes a difference.

7. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by irv Jan 31, 2021
• 1446 views

Then, where does the sh: 1: come from?

One way to know if bash shell is involved is to do a ${?} immediately after the failure. If you get the same error repeated, it happened in a shell script. (You enter all 4 characters, the dollar sign shown above is not the prompt)

Anyway, here's another puzzle for you:

? system_exec("ls ",0) 
puts(1,"done\n") 
while 1 do end while 

When run by Euphoria, this lists the files in the current directory, then prints a 0, followed by "done", and waits for a ctl-c

When run by Phix, this does nothing at all, just waits for a ctl-c and returns to the prompt. No file listing, no "done".

9. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by petelomax Jan 31, 2021
• 1412 views

? system_exec("ls ",0) 
puts(1,"done\n") 
while 1 do end while 

It has driven the compiler into an infinite loop.
In pEmit2.e, procedure scanforShortJmp(), line 723 the branch straightening checks to see if it is jumping to an unconditional jump and if it is, it hops direct to that target, which it also checks.
However if the offset on an unconditional jump is zero so it is jumping to itself, it will "follow that chain" forever:

                c2 = s5[c2]+5 
                offset += c2        -- ;- ) 
                s5[i3] = offset 
--31/1/21 
if c2=0 then exit end if 
            end while 

Alternatively, replace the offending loop(s):

--while 1 do end while 
while 1 do sleep(0.1) end while 
-- or 
{} = wait_key() 

11. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by petelomax Jan 31, 2021
• 1405 views

tom@tom-ideacentre-Y900-34ISZ:/media/fred/Seagate Backup Plus Drive/phix$ ./p64 -c hello 
sh: 1: /media/fred/Seagate: not found 

Ah, it is a "spaces in the path" problem, reproduced here by renaming my phix directory as "phix space":

pete@pete-VirtualBox:~/phix space$ ./p hello 
"hello" 
pete@pete-VirtualBox:~/phix space$ ./p -c hello 
sh: 1: /home/pete/phix: not found 

        if bind then 
>           string paramstr = "" 
            for i=1 to length(params) do 
--if 0 then 
--              outfile &= ' '&params[i] 
--else 
                pi = params[i] 
                if find(' ',pi) then 
                    pi = '\"'&pi&'\"' 
                end if 
<--             outfile &= ' '&pi 
>               paramstr &= ' '&pi 
--end if 
            end for 
            if bind>1 then 
<--             printf(1,"==>system_wait(%s):\n",{mainpath&outfile}) 
>               printf(1,"==>system_wait(%s):\n",{"\""&mainpath&outfile&"\""&paramstr}) 
            end if 
<--         errorcode = system_wait(mainpath&outfile) 
>           errorcode = system_wait("\""&mainpath&outfile&"\""&paramstr) 
            if bind=2 then 
                if errorcode=0 then 
<--                 outfile = mainpath&pnew2()&" -pmkr3" 
>                   outfile = "\""&mainpath&pnew2()&"\" -pmkr3" 
                    printf(1,"==>system_wait(%s):\n",{outfile}) 
                    errorcode = system_wait(outfile) 
                    if errorcode = 0 then 
<--                     outfile = mainpath&pnew()&" -pmkr4" 
>                       outfile = "\""&mainpath&pnew()&"\" -pmkr4" 
                        printf(1,"==>system_wait(%s):\n",{outfile}) 
                        errorcode = system_wait(outfile) 
                        if errorcode = 0 then 
                            printf(1,"self-host rounds 1..4 successfully completed (%3.2fs)\n",time()-t0) 

The < and > in column 1 are from a compare of my (windows) master and the fix I did on linux, delete them.
Instead of the usual ./p -c p, I had to use ./p p -c p to get it to recompile, that is in a directory with spaces in it.

13. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by irv Jan 31, 2021
• 1410 views

Just to be sure everything is equal before trying the detailed tests of phix on an external drive, I decided to remove all things phix from my computer and install a fresh copy, to make sure all the patches were up to date.

Testing[43]: /home/irv/phix/test/t43tchk.exw 
Testing[44]: /home/irv/phix/test/t44silly.exw 
{1,2,3,4,5} 
Testing[45]: /home/irv/phix/test/t45aod.exw 
started 
finished 
Testing[46]: /home/irv/phix/test/t46ltype.exw 
 
**NB** (reminder to self) Do *NOT* assume this is an error; 
 
 
/home/irv/phix/test/t46ltype.exw:610 
..teger} 
^ fail: ltype is 0b1111(object), not 0b0001(integer) (pilx86, pc=1084, vno=1206) 

Does that NB also apply to me? From here, it looks like something bad happened, but the following works properly.

cd phix 
./p hello 

Speaking of notes: ./p -c p does the "compile myself 4 times" stuff, and ends with "$ Overwriting p" and a blinking cursor which never goes away. After a short nap and a fresh cup of coffee, I grew tired of waiting and hit <enter>.

15. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by petelomax Jan 31, 2021
• 1429 views

Testing[43]: /home/irv/phix/test/t43tchk.exw 
Testing[44]: /home/irv/phix/test/t44silly.exw 
{1,2,3,4,5} 
Testing[45]: /home/irv/phix/test/t45aod.exw 
started 
finished 
Testing[46]: /home/irv/phix/test/t46ltype.exw 
 
**NB** (reminder to self) Do *NOT* assume this is an error; 
 
 
/home/irv/phix/test/t46ltype.exw:610 
..teger} 
^ fail: ltype is 0b1111(object), not 0b0001(integer) (pilx86, pc=1084, vno=1206) 

cd phix 
./p hello 

I "resurrected" that test on the 17th December, post-0.8.2

C:\Program Files (x86)\Phix\p.exw:295 --DEV p64 -c t46 is broken (put back in a broken state 17/12/20) 

--DEV fail 17/12/20: 
--      --/**/ #istype{o,integer} 

and at the top

--"DEV fail 17/12/20:" added in abundance! 

Of course I'd like localtypes to work better than they do, but I have to accept that there I wrote some code which is too clever for me to debug.
A "wrong" localtype (eg compiler thinks it is an integer when in fact it is a string) is utterly catastrophic, whereas
a "failed" localtype (eg compiler just says object when it could ideally say integer) just means slightly less efficient code will be emitted.
The NB is directed at me, a reflection/reminder of the fact there is an ok wrong as well as a not ok wrong, iyswim.

That "overwriting p" has always been like that, on linux, bit out of my hands really, I think I'll just not bother outputting it, on linux. [done]

17. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by irv Jan 31, 2021
• 1389 views

I guess I still misunderstand what is going on.

• I wiped all traces of Phix from my computer.
• ran the Linux 64-bit install script (installing from the web to /home/irv/phix)
• tests passed (with a note which Pete says we can safely ignore)
• $ Overwriting p...

? "Hello World" -- this is my program, saved in /home/irv/phix as hello.ex 

irv@irv-Mint19:~$ cd /phix 
irv@irv-Mint19:~/phix$ p 
 
Phix hybrid interpreter/compiler. 
Version 0.8.2 (64 bit Linux) Copyright Pete Lomax 2006..2016 
Enter ? for options or filename to execute: hello 
hello 
"Hello World" 

All good, except one might quibble about repeating the program name before activating it.

• ./p64 hello => bash: ./p64: No such file or directory

I don't understand why or how ./p64.exu could work to run anything, it's just two lines.

format ELF64 
include p.exw 

As far as I can tell, there is no p64 that is executable and no need to rename anything.

Let's clear that up before I tell you the really bad news. (Pete won't like this)

19. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by irv Jan 31, 2021
• 1363 views

OK, but why or how is _tom running ./p64 ? If in fact "mv p64 phix/p" succeeded, then trying to run ./p64 would result in an error (./p64: No such file or directory). Since it does (sometimes) work for him, that indicates that the "mv p64 phix/p" failed. In that case, one might wonder what else may have also failed.

The update completely destroyed object inheritance. Example: a has a property called "border_width"; private integer border_width; and set_border_width() are implemented in .e (via .e)

Window win = new() 
  win.border_width = 10 

Yesterday, all is fine. win.border_width = 10 worked. Today, I get a message: attempt to modify private field (border_width)

Worse, if I make the field public, it says: Segmentation fault (core dumped)

Worse than that, sometimes the error is: Cannot open /home/irv/pgtk/GtkContainer.ex

Say What? There is no GtkContainer.ex, and none of my code tries to include an *.ex file. A "grep *" of the entire directory shows:

GtkBin.e:5:include GtkContainer.e 
GtkBox.e:4:include GtkContainer.e 
GtkContainer.e:2:-- GtkContainer  

21. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by irv Jan 31, 2021
• 1338 views

Would that work? Aren't many of the latest fixes made in /builtins/VM files?

23. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by petelomax Feb 01, 2021
• 1288 views

I have uploaded a completely untested one for you, get it using wget http://phix.x10.mx/p64t (note the t for tom on the end)

I've also started a new site (as a backup/with https/edit playground, not planning on getting rid of the existing one) and uploaded it there: wget https://phix.heliohost.us/p64t

PS irv has a good point, there are 10 weeks worth of potential incompatibilites waiting in that file

25. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by petelomax Feb 01, 2021
• 1267 views

See https://openeuphoria.org/forum/135193.wc#135193

27. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by petelomax Feb 01, 2021
• 1264 views

Oh boy, are you not going to like this...

Unfortunately I have never been able to make it work that way. For instance, the recent addition of still_has_delete_routine():
1) create the routine, without using it. Rebuild.
2) use it (in the compiler itself). Rebuild.

If I try to merge those into one step, which is the same as updating the repo sources w/o a new release, it won't work anyway.
As a consequence I'm always very selective (/tardy) with repo updates.

I guess it's time for a new release.

29. Re: sh: 1: /media/mycomputer/Seagate: not found

• Posted by _tom (admin) Feb 01, 2021
• 1258 views

Success.

 wget https://phix.heliohost.us/p64t  

I also happen to have fresh files downloaded from github.

The -c switch now works "as advertised" on Linux.

It is interesting that this thread would never have happened if I had formatted my Seagate drive from Microsoft ntfs to a Linux drive--like I usually do.

thanks,

be well
_tom