Re: [bug-gawk] `exit' statement issue

[Denis Shirokov]

Hello Eli,


 @SET  EGAWK_PATH=\\CPU\CPU\DEV\PROJECT\eGAWK\

It's the builder path(see below) - it's not used by .GWK files. can be dropped

  @SET  EGAWK_CMDLINE= %*

Through envar EGAWK_CMDLINE the entire commandline string is available
inside the script by reading it from ENVIRON["EGAWK_CMDLINE"]. In our
case this doesn't matter

non-ASCII characters:

I really believe that this not the point.
There is a two cases when I use non-ASC2:

chars: x00-x1F:
This characters can't be contained inside .GWK files(the file that I
was sending to you while reporting `exit' statement). Because .GWK
files is generated by the Builder.
I use this characters (actually only three of them: x01 and
x02-x03(pairly)) as the prefixes in the source.
The prefixes that preceeding special builder's instructions.

Example:
x01.use      <libname,...>   - is including some libraries
or
x02<macroname>x03                  - is calling L0-macros

Anyway, this characters do not pass to the final build because
together with the text of that kind builder special instructions - it
is totally excluded from the source.
If it isn't then please report me what characters from range x00-x1F
you are seeing and where(of course excluding x09 and x0D,x0A =)

other non-ASCII chars:

I think that as like the some indecent names in my script just will be
better to ingore it. I always use code page 1252 and before start to
use it - i was totally testing with the GAWK all characters of this
encoding : in the names, expressions, strings, regexps, commentaries
and so on. I think that because source that i was sending to you was
been just current temporary source where i just testing some of the
new functions - such kind of characters may be found inside it. As
like the indecent names =) Sorry about it)


Please note that yesterday i was sending You, Arnold and also Manuel -
report about finding `exit' statement issue reason.

If rate severity(the size of the needed job to fix the issue) from 0
to 5: where 0 is the easiest one and 5 is requiring most bigger
change: then fixing issue with the `exit' statement i rate as
1.(please remember that actually I know nothing about GAWK code)

I think that ESP should be stored at the begining of the every first
expression, and `exit' should used it.)

Thank You and Sorry for Inconviniences

Denis

PS. I found a several ways to hardly crash GAWK 4.1 while working with
FUNCTAB. Of course if using `delete' statement with this array =)
Actually this array may be more useful than just for detection - is
function available or not inside the script. But then it's needs to be
extended.
Are you, guys, interested about reports of GAWK 4.1 crashing while
using `delete' statement with FUNCTAB array?


2013/6/1, Eli Zaretskii <address@hidden>:
>> Date: Sat, 01 Jun 2013 15:15:25 +0300
>> From: Eli Zaretskii <address@hidden>
>> Cc: address@hidden, address@hidden
>>
>> > From: Aharon Robbins <address@hidden>
>> > Date: Tue, 28 May 2013 22:37:38 +0300
>> >
>> > Since your script runs only on Windoew, I cannot test it on Linux,
>> > perhaps Eli will try.
>>
>> Just tried, and I can run it on GNU/Linux with no trouble, and I get
>> the same error message at exit.  I edited the _fatal function to this:
>>
>>   func _fatal(t) {
>>        exit }
>>
>> and also invoked the script directly, like this:
>>
>>   gawk -f _main.gwk _main.gwk
>>
>> and it still exits with the same fatal error message.  I sent in a
>> previous message the data I was able to collect from the fatal error.
>
> Note that the reformatted script you (Aharon) sent does not include
> any non-ASCII characters, while the original one did.  Perhaps those
> characters are somehow related to the fatal eror.
>
> Denis, what encoding was that script in, and what was the purpose of
> using those non-ASCII characters?
>