[Oorexx-devel] Subtle gcc-error on Apple
Hi there, a couple of weeks ago, after changing the code of BSF4ooRexx, the Apple version started to behave erroneously. Originally, I thought that I introduced some error into the code and tried to debug this whenever I got some free time. It turns out, that there is a subtle gcc-error on Apple (using i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5664)): if the source code employs inline functions it may be the case that a global variable pointing to a structure may do so wrongly! In my case I could remedy the situation - after having narrowed the problem down to this rather exotic phenomenon - by just removing the inline keyword from the Apple compilation! Thought that those of you who are working on Apple should be aware of this bug. However, in this particular case the source code of BSF4ooRexx.cc is not yet cleaned which means it contains an unbelievable amount of defs and inlines from the past ten years, so this complexity may be needed for getting into this situation (if the preprocessor is causing the bug). Regards, ---rony -- Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d ___ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel
Re: [Oorexx-devel] Question ad DecodeConditionInfo(...)
Sorry, after changing the code, it just occurred to me: cond.program
is a RexxStringObject and should never take on NULL as Jean-Louis
reports, but should rather be Nil()?
Or with other words: is it possible that the RexxStringObject members in
a RexxCondition structure can be NULL?
---rony
On 19.03.2011 20:27, Rony G. Flatscher wrote:
Sorry, it should read Nil in the text rather than NULL.
---rony
On 19.03.2011 20:12, Rony G. Flatscher wrote:
In a bug report which relates to BSF4ooRexx Jean-Louis was kind
enough to look into the issue and came up with an analysis in
https://sourceforge.net/tracker/?func=detailatid=684730aid=3205263group_id=119701.
According to his findings, the RexxCondition member program was
NULL in the supplied example. Before changing/adapting the code, I
was wondering whether it is truly possible for the program member
to be NULL in the case of a SYNTAX condition?
What about the members errortext, code, message, could they be
NULL as well in SYNTAX condition?
Is it safe to assume that the member conditionName will always have
a string value?
---rony
P.S.: Here is the code that uses DecodeConditionInfo(...) to create a
Rexx-like error message which then gets forwarded to the Java side,
where the line causing the crash is highlighted in bold red.
// allocates and creates a Rexx-like error message
inline char * RgfCreateRexxlikeErrorInfo (RexxThreadContext *rtc,
RexxDirectoryObject condObj, const char * header)
{
rtc-ClearCondition();
RexxCondition cond;
rtc-DecodeConditionInfo(condObj, cond);
char *msg=(char *)RexxAllocateMemory( 4096 );
msg[0]=0;
// format message according to exception type
if (rtc-SendMessage1(cond.conditionName, EQUALS,
rtc-String(SYNTAX))==rtc-True()) // SYNTAX condition ?
{
// will only use 1024 of it
char const
*line1=rtc-CString(rtc-SendMessage0(rtc-DirectoryAt(condObj,
TRACEBACK), FIRSTITEM)) ; // first traceback line=error line
char line2[2048]=;// will only use 1024 of it
SNPRINTF(line2, 2048, Error %d running %.512s line %d:
%.1024s,
* (int) cond.rc,
rtc-CString(cond.program),*
(int) cond.position,
rtc-CString(cond.errortext) );
char line3[2048]=;// will only use 1024 of it
SNPRINTF( line3, 2048, Error %d.%d: %.1024s, (int) (cond.code
/ 1000),
(int) (cond.code % 1000),
rtc-CString(cond.message));
SNPRINTF( msg, 4096, %.16s%.512s:\n%.1024s\n%.1024s\n%.1024s,
DLLNAME, header, line1, line2, line3);
}
else
{
SNPRINTF( msg, 4096, %.16s%.512s: Rexx condition [%.512s]
raised, DLLNAME, header, rtc-CString(cond.conditionName));
}
return msg;
}
--
Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d___
Oorexx-devel mailing list
Oorexx-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oorexx-devel
[Oorexx-devel] Beta-test for new version of BSF4ooRexx ...
Hi there, at http://wi.wu.ac.at/rgf/rexx/bsf4oorexx/archive/2011/2011-03-19/ there is a new BSF4ooRexx package, which is at a beta level (needs testing on the various platforms). It should fix the problem reported by Uli Zinngrebe. Will be testing this version over this weekend and once regarded to be o.k. I will move on to setup an own BSF4ooRexx communication area on Sourceforge (probably next weekend), such that purely BSF4ooRexx related discussions can be moved there. ---rony -- Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d ___ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel
Re: [Oorexx-devel] Question ad DecodeConditionInfo(...)
Why do you clear the condition before you you decode it? Without looking at
code, can't say for sure, but that seems risky. How do you know clearing
the condition doesn't set program to null?
On Mar 19, 2011 12:12 PM, Rony G. Flatscher rony.flatsc...@wu-wien.ac.at
wrote:
In a bug report which relates to BSF4ooRexx Jean-Louis was kind enough to
look into the issue and came up with an analysis in
https://sourceforge.net/tracker/?func=detailatid=684730aid=3205263group_id=119701https://sourceforge.net/tracker/?func=detailatid=684730aid=3205263group_id=119701
.
According to his findings, the RexxCondition member program was NULL in
the supplied example. Before changing/adapting the code, I was wondering
whether it is truly possible for the program member to be NULL in the case
of a SYNTAX condition?
What about the members errortext, code, message, could they be NULL as
well in SYNTAX condition?
Is it safe to assume that the member conditionName will always have a
string value?
---rony
P.S.: Here is the code that uses DecodeConditionInfo(...) to create a
Rexx-like error message which then gets forwarded to the Java side, where
the line causing the crash is highlighted in bold red.
// allocates and creates a Rexx-like error message
inline char * RgfCreateRexxlikeErrorInfo (RexxThreadContext *rtc,
RexxDirectoryObject condObj, const char * header)
{
rtc-ClearCondition();
RexxCondition cond;
rtc-DecodeConditionInfo(condObj, cond);
char *msg=(char *)RexxAllocateMemory( 4096 );
msg[0]=0;
// format message according to exception type
if (rtc-SendMessage1(cond.conditionName, EQUALS,
rtc-String(SYNTAX))==rtc-True()) // SYNTAX condition ?
{
// will only use 1024 of it
char const
*line1=rtc-CString(rtc-SendMessage0(rtc-DirectoryAt(condObj,
TRACEBACK), FIRSTITEM)) ; // first traceback line=error line
char line2[2048]=;// will only use 1024 of it
SNPRINTF(line2, 2048, Error %d running %.512s line %d: %.1024s,
* (int) cond.rc, rtc-CString(cond.program),*
(int) cond.position, rtc-CString(cond.errortext) );
char line3[2048]=;// will only use 1024 of it
SNPRINTF( line3, 2048, Error %d.%d: %.1024s, (int)
(cond.code / 1000),
(int) (cond.code % 1000),
rtc-CString(cond.message));
SNPRINTF( msg, 4096,
%.16s%.512s:\n%.1024s\n%.1024s\n%.1024s, DLLNAME, header, line1,
line2, line3);
}
else
{
SNPRINTF( msg, 4096, %.16s%.512s: Rexx condition [%.512s]
raised, DLLNAME, header, rtc-CString(cond.conditionName));
}
return msg;
}
--
Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d
___
Oorexx-devel mailing list
Oorexx-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oorexx-devel
--
Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d___
Oorexx-devel mailing list
Oorexx-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oorexx-devel
Re: [Oorexx-devel] CPL V1 License superseded
Thanks Rick On 19 March 2011 21:46, Rick McGuire object.r...@gmail.com wrote: No. The cpl is just as valid as it ever was. It doesn't matter. Rick Sent from my iPad On Mar 19, 2011, at 5:27 PM, Sahananda (Jon) Wolfers sahana...@windhorse.biz wrote: Hi All, I don't know if this matters, but I notice herehttp://www.opensource.org/licenses/cpl1.0.phpthat the CPL license has been superseded by the Eclipse Public License. Does this matter? Does it mean ooRexx will migrate? thanks, Jon -- Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d ___ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel -- Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d ___ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel -- Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d___ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel
Re: [Oorexx-devel] Question ad DecodeConditionInfo(...)
Sent from my iPad
On Mar 19, 2011, at 3:41 PM, Rony G. Flatscher rony.flatsc...@wu-wien.ac.at
wrote:
Sorry, after changing the code, it just occurred to me: cond.program is a
RexxStringObject and should never take on NULL as Jean-Louis reports, but
should rather be Nil()?
They most certainly can be null. A null value indicates an absence of a value.
A RexxStringObject can
Never be .nil, since that would violate the type guarantee.
This is the appropriate value here. The action of raise propagate is to use
the current condition
Information and reraise the condition in the caller's context as if the
condition had been triggered at
The point of the invoking call. Since the calling context in this situation is
the bsf4rexx native code,
There is no program information available.
Rick
Or with other words: is it possible that the RexxStringObject members in a
RexxCondition structure can be NULL?
---rony
On 19.03.2011 20:27, Rony G. Flatscher wrote:
Sorry, it should read Nil in the text rather than NULL.
---rony
On 19.03.2011 20:12, Rony G. Flatscher wrote:
In a bug report which relates to BSF4ooRexx Jean-Louis was kind enough to
look into the issue and came up with an analysis in
https://sourceforge.net/tracker/?func=detailatid=684730aid=3205263group_id=119701.
According to his findings, the RexxCondition member program was NULL in
the supplied example. Before changing/adapting the code, I was wondering
whether it is truly possible for the program member to be NULL in the
case of a SYNTAX condition?
What about the members errortext, code, message, could they be NULL
as well in SYNTAX condition?
Is it safe to assume that the member conditionName will always have a
string value?
---rony
P.S.: Here is the code that uses DecodeConditionInfo(...) to create a
Rexx-like error message which then gets forwarded to the Java side, where
the line causing the crash is highlighted in bold red.
// allocates and creates a Rexx-like error message
inline char * RgfCreateRexxlikeErrorInfo (RexxThreadContext *rtc,
RexxDirectoryObject condObj, const char * header)
{
rtc-ClearCondition();
RexxCondition cond;
rtc-DecodeConditionInfo(condObj, cond);
char *msg=(char *)RexxAllocateMemory( 4096 );
msg[0]=0;
// format message according to exception type
if (rtc-SendMessage1(cond.conditionName, EQUALS,
rtc-String(SYNTAX))==rtc-True()) // SYNTAX condition ?
{
// will only use 1024 of it
char const
*line1=rtc-CString(rtc-SendMessage0(rtc-DirectoryAt(condObj,
TRACEBACK), FIRSTITEM)) ; // first traceback line=error line
char line2[2048]=;// will only use 1024 of it
SNPRINTF(line2, 2048, Error %d running %.512s line %d: %.1024s,
(int) cond.rc, rtc-CString(cond.program),
(int) cond.position,
rtc-CString(cond.errortext) );
char line3[2048]=;// will only use 1024 of it
SNPRINTF( line3, 2048, Error %d.%d: %.1024s, (int) (cond.code /
1000),
(int) (cond.code % 1000),
rtc-CString(cond.message));
SNPRINTF( msg, 4096, %.16s%.512s:\n%.1024s\n%.1024s\n%.1024s,
DLLNAME, header, line1, line2, line3);
}
else
{
SNPRINTF( msg, 4096, %.16s%.512s: Rexx condition [%.512s] raised,
DLLNAME, header, rtc-CString(cond.conditionName));
}
return msg;
}
--
Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d
___
Oorexx-devel mailing list
Oorexx-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oorexx-devel
--
Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d___
Oorexx-devel mailing list
Oorexx-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oorexx-devel
Re: [Oorexx-devel] Question ad DecodeConditionInfo(...)
On 19.03.2011 22:52, Rick McGuire wrote: Sent from my iPad On Mar 19, 2011, at 3:41 PM, Rony G. Flatscher rony.flatsc...@wu-wien.ac.at mailto:rony.flatsc...@wu-wien.ac.at wrote: Sorry, after changing the code, it just occurred to me: cond.program is a RexxStringObject and should never take on NULL as Jean-Louis reports, but should rather be Nil()? They most certainly can be null. A null value indicates an absence of a value. A RexxStringObject can Never be .nil, since that would violate the type guarantee. This is the appropriate value here. The action of raise propagate is to use the current condition Information and reraise the condition in the caller's context as if the condition had been triggered at The point of the invoking call. Since the calling context in this situation is the bsf4rexx native code, There is no program information available. Thank you very much for clarifying this, Rick! ---rony -- Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d___ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel
