Hi,

my goal is to transfer the "errorInfo" into a "stacktrace" used by my SW.

-> I require: command, fileName and lineNumber
-> the [info errorstack] return "something" which is missing some essential parts.

[info errorstack]:
> INNER {invokeStk1 ::tcllcconfig::LcSettingC::MK_NULL LookupBool in} CALL {::oo::Obj40::my LcSettingLookupBool}

as I understand:

$setting LookupBool $name == ::tcllcconfig::LcSettingC::MK_NULL LookupBool in
-> problem: post versa pre parse expression
-> problem: the fileName(class etc) is *not* available

==== lcconfig-2-0-5-(0-0-0+binary+pipe+tcl.pipe.pipe) use invalid cfs FAILED
---- Result was:
C> (client-0-0-0) [ServiceCallError] <Tok|2E3A471B{LcSettingRpc::LookupBool}> <Num|1>
S> (server-0-0-0) [tcl_mqmsgque_ServiceCall] 'LcSettingC' hdl is NULL
while executing
"$setting LookupBool $name"
(class "::LibLcConfigRpcServer" method "LcSettingLookupBool" line 4)
invoked from within
"::oo::Obj40::my LcSettingLookupBool"
C> (client-0-0-0) | 871 | send_mq.c | MqSendEND_AND_WAIT_RT
C> (client-0-0-0) | 508 | high_lng.h | $ctx Send 'W' ... '2E3A471B{LcSettingRpc::LookupBool}:HC@O' ...
C> (client-0-0-0) | 522 | high_lng.h | $ctx Send...
C> (client-0-0-0) | 75 | ~KernelRpcClient.tcl | my Send $typ $fmt {*}$args
C> (client-0-0-0) | 588 | ~ConfigRpcClient.tcl | $rpc send "W" "2E3A471B{LcSettingRpc::LookupBool}:HC@O" $hdl $name
C> (client-0-0-0) | 3 | lcconfig.test | $null LookupBool in
C> (client-0-0-0) | 2405 | tcltest-2.5.3.tm | uplevel 1 $script
C> (client-0-0-0) | 1 | lcconfig.test | ::tcltest::RunTest lcconfig-2-0-5-(0-0-0+binary+pipe+tcl.pipe.pipe) { ...
C> (client-0-0-0) | 1651 | tcltest-2.5.3.tm | uplevel 1 $script
C> (client-0-0-0) | 1 | lcconfig.test | ::tcltest::Eval {::tcltest::RunTest
lcconfig-2-0-5-(0-0-0+binary+pipe+tcl.pipe.pipe) { ...
C> (client-0-0-0) | 2021 | tcltest-2.5.3.tm | uplevel 1 [list [namespace origin Eval] $command 1]
C> (client-0-0-0) | 87 | lcconfig.test | test lcconfig-2-0-5-($F+$B+$C+$S) { use invalid cfs } -constraints {$B
eq "binary"} -setup { ...
C> (client-0-0-0) | 6 | lcconfig.test | foreach F [handles] { ...
C> (client-0-0-0) | 649 | library.tcl | uplevel $code
C> (client-0-0-0) | 195 | lcconfig.test | Block 2 { ...
C> (client-0-0-0) | 104 | lcconfig.test | foreach S [filter SRV_LST $C ] { ...
C> (client-0-0-0) | 101 | lcconfig.test | foreach C [getEnv COM_LST] { ...
C> (client-0-0-0) | 98 | lcconfig.test | foreach B [getEnv BIN_LST] { ...

Hi,
remark that "info errorstack" and the stack trace are different things.
In the first, variable names are replaced by values.
In the second, there are the variable names "$var".

They don't match directly.

THanks,
Harald

Am 25.11.2024 um 22:13 schrieb aotto1968:
> Hi,
>
> my goal is to transfer the "errorInfo" into a "stacktrace" used by my SW.
>
> -> I require: command, fileName and lineNumber
> -> the [info errorstack] return "something" which is missing some
> essential parts.
>
> [info errorstack]:
> > INNER {invokeStk1 ::tcllcconfig::LcSettingC::MK_NULL LookupBool in}
> CALL {::oo::Obj40::my LcSettingLookupBool}
>
> as I understand:
>
> $setting LookupBool $name == ::tcllcconfig::LcSettingC::MK_NULL
> LookupBool in
> -> problem: post versa pre parse expression
> -> problem: the fileName(class etc) is *not* available
>
>
> ==== lcconfig-2-0-5-(0-0-0+binary+pipe+tcl.pipe.pipe) use invalid cfs
> FAILED
> ---- Result was:
> C> (client-0-0-0)   [ServiceCallError] <Tok|
> 2E3A471B{LcSettingRpc::LookupBool}> <Num|1>
> S> (server-0-0-0)   [tcl_mqmsgque_ServiceCall] 'LcSettingC' hdl is NULL
>     while executing
> "$setting LookupBool $name"
>     (class "::LibLcConfigRpcServer" method "LcSettingLookupBool" line 4)
>     invoked from within
> "::oo::Obj40::my LcSettingLookupBool"
> C> (client-0-0-0)     | 871   | send_mq.c            |
> MqSendEND_AND_WAIT_RT
> C> (client-0-0-0)     | 508   | high_lng.h           | $ctx Send 'W' ...
> '2E3A471B{LcSettingRpc::LookupBool}:HC@O' ...
> C> (client-0-0-0)     | 522   | high_lng.h           | $ctx Send...
> C> (client-0-0-0)     | 75    | ~KernelRpcClient.tcl | my Send $typ $fmt
> {*}$args
> C> (client-0-0-0)     | 588   | ~ConfigRpcClient.tcl | $rpc send "W"
> "2E3A471B{LcSettingRpc::LookupBool}:HC@O" $hdl $name
> C> (client-0-0-0)     | 3     | lcconfig.test        | $null LookupBool in
> C> (client-0-0-0)     | 2405  | tcltest-2.5.3.tm     | uplevel 1 $script
> C> (client-0-0-0)     | 1     | lcconfig.test
> | ::tcltest::RunTest lcconfig-2-0-5-(0-0-0+binary+pipe+tcl.pipe.pipe) { ...
> C> (client-0-0-0)     | 1651  | tcltest-2.5.3.tm     | uplevel 1 $script
> C> (client-0-0-0)     | 1     | lcconfig.test        | ::tcltest::Eval
> {::tcltest::RunTest lcconfig-2-0-5-(0-0-0+binary+pipe+tcl.pipe.pipe) { ...
> C> (client-0-0-0)     | 2021  | tcltest-2.5.3.tm     | uplevel 1 [list
> [namespace origin Eval] $command 1]
> C> (client-0-0-0)     | 87    | lcconfig.test        | test
> lcconfig-2-0-5-($F+$B+$C+$S) { use invalid cfs }  -constraints {$B eq
> "binary"}  -setup { ...
> C> (client-0-0-0)     | 6     | lcconfig.test        | foreach F
> [handles] { ...
> C> (client-0-0-0)     | 649   | library.tcl          | uplevel $code
> C> (client-0-0-0)     | 195   | lcconfig.test        | Block 2 { ...
> C> (client-0-0-0)     | 104   | lcconfig.test        | foreach S [filter
> SRV_LST $C ] { ...
> C> (client-0-0-0)     | 101   | lcconfig.test        | foreach C [getEnv
> COM_LST] { ...
> C> (client-0-0-0)     | 98    | lcconfig.test        | foreach B [getEnv
> BIN_LST] { ...

On 26.11.24 09:09, Harald Oehlmann wrote:
> Hi,
> remark that "info errorstack" and the stack trace are different things.
> In the first, variable names are replaced by values.
> In the second, there are the variable names "$var".
>
> They don't match directly.
>
> THanks,
> Harald
>

This is funny because "errorInfo" and "errorStack" are written in the same proc "

/*
*----------------------------------------------------------------------
*
* TclLogCommandInfo --
*
* This function is invoked after an error occurs in an interpreter. It
* adds information to iPtr->errorInfo/errorStack fields to describe the
* command that was being executed when the error occurred. When pc and
* tosPtr are non-NULL, conveying a bytecode execution "inner context",
* and the offending instruction is suitable, that inner context is
* recorded in errorStack.
*
* Results:
* None.
*
* Side effects:
* Information about the command is added to errorInfo/errorStack and the
* line number stored internally in the interpreter is set.
*
*----------------------------------------------------------------------
*/

and the "errorInfo" has a BETTER info than the "errorStack" ?

I use to raise MY error in a my-error-raise-proc and I use [info frame..] to collect all
the required data to build my-errorStack proper.

The problem is the "errorStack" from OTHER software or from TCL self witch do NOT
collect all the available Info.

My question is simple:

why someone program a [info errorstack ...] to provide an API for a missing feature and
do NOT call all the data available from [info frame] ?

TCL could have a wonderful "exception" interface

1. The (Tcl_GetReturnOptions) dict is the "exception-object"
> example:
> Tcl_GetReturnOptions(interp,TCL_ERROR): -code 1 -level 0 -errorstack {INNER {invokeStk1
::tcllcconfig::LcSettingC::MK_NULL LookupBool in} CALL {::oo::Obj40::my LcSettingLookupBool}} -errorcode NONE -errorinfo
{'LcSettingC' hdl is NULL
> while executing
> "$setting LookupBool $name"
> (class "::LibLcConfigRpcServer" method "LcSettingLookupBool" line 4)
> invoked from within
> "::oo::Obj40::my LcSettingLookupBool"} -errorline 1
2. The "-errorstack" key has the LIST of [info frame] dicts until catch

-> everyone is happy

* aotto1968 <aotto1968@t-online.de>
| TCL could have a wonderful "exception" interface
>
| 1. The (Tcl_GetReturnOptions) dict is the "exception-object"
| > example:
| > Tcl_GetReturnOptions(interp,TCL_ERROR): -code 1 -level 0
| > -errorstack {INNER {invokeStk1
| > ::tcllcconfig::LcSettingC::MK_NULL LookupBool in} CALL
| > {::oo::Obj40::my LcSettingLookupBool}} -errorcode NONE
| > -errorinfo
| {'LcSettingC' hdl is NULL
| > while executing
| > "$setting LookupBool $name"
| > (class "::LibLcConfigRpcServer" method "LcSettingLookupBool" line 4)
| > invoked from within
| > "::oo::Obj40::my LcSettingLookupBool"} -errorline 1
| 2. The "-errorstack" key has the LIST of [info frame] dicts until catch
>
| -> everyone is happy

I guess your contribution to the TCL code base would probably be
welcome, provided there is agreement on the necessity of this
feature...

=> https://core.tcl-lang.org/tips/doc/trunk/doc/help.md

R'

On 27.11.24 10:29, Ralf Fassel wrote:
> * aotto1968 <aotto1968@t-online.de>
> | TCL could have a wonderful "exception" interface
>>
> | 1. The (Tcl_GetReturnOptions) dict is the "exception-object"
> | > example:
> | > Tcl_GetReturnOptions(interp,TCL_ERROR): -code 1 -level 0
> | > -errorstack {INNER {invokeStk1
> | > ::tcllcconfig::LcSettingC::MK_NULL LookupBool in} CALL
> | > {::oo::Obj40::my LcSettingLookupBool}} -errorcode NONE
> | > -errorinfo
> | {'LcSettingC' hdl is NULL
> | > while executing
> | > "$setting LookupBool $name"
> | > (class "::LibLcConfigRpcServer" method "LcSettingLookupBool" line 4)
> | > invoked from within
> | > "::oo::Obj40::my LcSettingLookupBool"} -errorline 1
> | 2. The "-errorstack" key has the LIST of [info frame] dicts until catch
>>
> | -> everyone is happy
>
> I guess your contribution to the TCL code base would probably be
> welcome, provided there is agreement on the necessity of this
> feature...
>
> => https://core.tcl-lang.org/tips/doc/trunk/doc/help.md
>
> R'

The problem is always "time" … I work with a dozens of languages and TCL is just one of them…
for example today I figured out how to implement this in python…

def MkErrorC_FormatException(err,exp,caller):
err.NoRaise().SetC(str(exp),caller)
tb = exp.__traceback__
while tb is not None:
err.StackFormat(
tb.tb_frame.f_code.co_qualname,
tb.tb_frame.f_code.co_filename,
tb.tb_lineno
)
tb = tb.tb_next

and by-the-way I add a feature to direct "inject" python-code into a python-c-extension without
shipping an extra python source file.

→ My problem is always NOT to drift-away into old-style-language-development…