on errorlevel causes btm to mess up TCC

[David Marcus]

I don't understand what is happening with the following btm file. Without the "on errorlevel" statement, it works fine.

C:\Junk>dir

Volume in drive C is OS Serial number is 1ce5:1203
Directory of C:\Junk\*

3/12/2011 6:04p DIR .
3/12/2011 6:04p DIR ..
3/12/2011 6:03p 256 Foo.btm
3/12/2011 5:54p 745,282 ZippedListDownload1.zip
745,538 bytes in 2 files and 2 dirs 749,568 bytes allocated
146,656,620,544 bytes free

C:\Junk>type Foo.btm
*on errormsg cancel
*on errorlevel ge 1 cancel
*setlocal
*unalias *
echo on

alias Extract=`iff %@filesize[%1.zip] gt 200 then %+ unzip /o /q %1.zip %+ else %+ %=
echo Missing or bad %1 file %+ endiff`

Extract ZippedListDownload1

C:\Junk>Foo.btm
alias Extract=`iff %@filesize[%1.zip] gt 200 then %+ unzip /o /q %1.zip %+ else %+ echo Missing or bad %1 file %+ endiff`
iff %@filesize[ZippedListDownload1.zip] gt 200 then
cancel
C:\Junk>dir

C:\Junk>cd \

C:\Junk>dir

C:\Junk>ver /r

C:\Junk>

At this point, the console ignores whatever I type and just returns a prompt. The result is the same if I run TCC inside TC or standalone.

TCC 12.01.44 Windows Vista [Version 6.0.6002]

 

[David Marcus]

Still happens with

TCC 12.10.56 x64 Windows 7 [Version 6.1.7601]

Is "on errorlevel" broken?

 

[rconn]

I don't understand what is happening with the following btm file. Without the "on errorlevel" statement, it works fine.

It doesn't have anything to do with ON ERRORLEVEL; it's the CANCEL inside an IFF statement that's confusing the parser. That's been there forever; I'll look at it but it's a dangerous place to change (lots of opportunity to break backwards compatibility).

 

[David Marcus]

Does that mean I shouldn't use CANCEL in ON ERRORMSG or ON ERRORLEVEL if the cancel might happen within an IFF block? Does the same apply to IF? Is QUIT also a problem? Is it OK to put CANCEL or QUIT within an IFF block or in an IF statement?

 

[rconn]

Does that mean I shouldn't use CANCEL in ON ERRORMSG or ON ERRORLEVEL if the cancel might happen within an IFF block? Does the same apply to IF? Is QUIT also a problem? Is it OK to put CANCEL or QUIT within an IFF block or in an IF statement?

The problem wasn't simply a CANCEL inside an IFF, it was a CANCEL called by an ON inside an IFF.

12.0.57 (uploaded a couple of hours ago) has a workaround.

 

[David Marcus]

I installed the new version and now the btm ends normally. But, I don't understand why the ON is firing. First without ON ERRORLEVEL:

C:\Junk>dir

 Volume in drive C is TI105444W0C    Serial number is 709c:de9d
 Directory of  C:\Junk\*

 3/24/2011  11:19a        DIR    .
 3/24/2011  11:19a        DIR    ..
 3/24/2011  11:19a            275  Foo.btm
 3/22/2011   3:42p        715,367  ZippedListDownload1.zip
           715,642 bytes in 2 files and 2 dirs    720,896 bytes allocated
   249,338,101,760 bytes free

C:\Junk>type Foo.btm
*on errormsg cancel
rem *on errorlevel ge 1 cancel
*setlocal
*unalias *
echo on

alias Extract=`iff %@filesize[%1.zip] gt 200 then %+ echo %? %_? %+ unzip /o /q %1.zip %+ echo %? %_? %+ else %+ %=
echo Missing or bad %1 file %+ endiff`

Extract ZippedListDownload1

C:\Junk>Foo.btm
alias Extract=`iff %@filesize[%1.zip] gt 200 then %+ echo %? %_? %+ unzip /o /q %1.zip %+ echo %? %_? %+ else %+ echo Missing or bad %1 file %+ endiff`
iff %@filesize[ZippedListDownload1.zip] gt 200 then
echo 0 0
0 0
unzip /o /q ZippedListDownload1.zip
echo 0 0
0 0

Now with ON ERRORLEVEL:

C:\Junk>type Foo.btm
*on errormsg cancel
*on errorlevel ge 1 cancel
*setlocal
*unalias *
echo on

alias Extract=`iff %@filesize[%1.zip] gt 200 then %+ echo %? %_? %+ unzip /o /q %1.zip %+ echo %? %_? %+ else %+ %=
echo Missing or bad %1 file %+ endiff`

Extract ZippedListDownload1

C:\Junk>Foo.btm
alias Extract=`iff %@filesize[%1.zip] gt 200 then %+ echo %? %_? %+ unzip /o /q %1.zip %+ echo %? %_? %+ else %+ echo Missing or bad %1 file %+ endiff`
iff %@filesize[ZippedListDownload1.zip] gt 200 then
cancel

C:\Junk>

Also, the Help for ERRORLEVEL in the first sentence says _? when it means ?.

 

[rconn]

I installed the new version and now the btm ends normally. But, I don't understand why the ON is firing.

The ON ERRORLEVEL is executing because the IFF is returning a non-zero value (required in order to allow you to pipe the output of an IFF, and to support the || and && conditional operators).

 

[David Marcus]

It seems the exit code isn't put in the variables:

C:\Junk>type Foo.btm
*on errorlevel ge 1 echo on errorlevel %errorlevel %? %_?
iff 300 gt 200 then
   echo greater
else
   echo less
endiff
echo %errorlevel %? %_?

C:\Junk>Foo.btm
on errorlevel 0 0 0
greater
0 0 0

Perhaps I'm missing something, but ON ERRORLEVEL doesn't seem too useful if various internal commands will trigger it, even when they aren't producing an error.