ikrabbe / tclkit Goto Github PK

trying to wrap a starpack with the latest sdx the same way as ever with my old 
sdx dated from 2008.
I see this error (I only show relevant lines):

Runtime: tclkit-gui-859.exe 
Sdx    : C:\Programme\MiscTool\sdx.exe
Projekt: menue
:
:
 customizing tclkit.ico in executable
  icon 64x64/32: replaced
  icon 48x48/32: replaced
  icon 32x32/32: replaced
  icon 24x24/32: replaced
  icon 16x16/32: replaced
  icon 48x48/24: replaced
  icon 32x32/24: replaced
  icon 24x24/24: replaced
  icon 16x16/24: replaced
  icon 48x48/8: replaced
  icon 32x32/8: replaced
  icon 24x24/8: replaced
  icon 16x16/8: replaced
 customizing strings in executable
  FileDescription: set to 'Anwendungsmenue'
  CompanyName: set to 'BMT'
  ProductName: set to 'menue'
  LegalCopyright: set to 'Copyright (c) M.Hoffmann'
  ProductVersion: set to '1.03ß'
195 updates applied
error deleting "menue/tclkit.ico": cross-domain link
    while executing
"file delete menue/tclkit.ico"
    ("eval" body line 1)
    invoked from within
"eval $args"
    invoked from within
"_rsync opts file delete $dest"
    (procedure "rcopy" line 24)
    invoked from within
"rcopy opts $file $target"
    (procedure "rsync" line 63)
    invoked from within
"rsync opts $src $dest"
    invoked from within
"set n [rsync opts $src $dest]"
    (file "C:/Programme/MiscTool/sdx.exe/lib/app-sdx/sync.tcl" line 317)
    invoked from within
"source [file join [file dirname [info script]] sync.tcl] "
    invoked from within
"if {![catch { package require Mk4tcl }]} {
  vfs::mk4::Mount $out $out
  set argv $syncopts
  lappend argv -compress $compress -verbose $verbose -noer..."
    (file "C:/Programme/MiscTool/sdx.exe/lib/app-sdx/wrap.tcl" line 272)
    invoked from within
"source C:/Programme/MiscTool/sdx.exe/lib/app-sdx/wrap.tcl"
    ("uplevel" body line 1)
    invoked from within
"uplevel #0 [list source [file join $b $a.tcl]]"
    (procedure "run_sdx" line 17)
    invoked from within
"run_sdx"
    (file "C:/Programme/MiscTool/sdx.exe/lib/app-sdx/sdx.tcl" line 59)
    invoked from within
"source C:/Programme/MiscTool/sdx.exe/lib/app-sdx/sdx.tcl"
    ("package ifneeded app-sdx 2.0" script)
    invoked from within
"package require app-sdx"
    (file "C:/Programme/MiscTool/sdx.exe/main.tcl" line 8)
:
What's goin' on here?

What steps will reproduce the problem?
1. On windows, run tclkit inside a mintty prompt (Cygwin environment)
2. try do execute interactive commands on prompt
3. execute command "console show"
4. try do execute interactive commands on console

Output from commands are not showing or are outputed on the wrong place. This 
problem doesn't happen if tclkit is started from windows console (cmd.exe).

What steps will reproduce the problem?
1. Create a starpack that outputs a CGI:
   #! /usr/bin/env tclsh
   puts "Content-type: text/plain\n\nHello World"
2. Run the Starpack CGI under Apache with an argument i.e.:
    http://example.com/cgi-bin/hello-world.cgi?/etc/passwd
3. Notice that the Starpack tried to execute the contents of "/etc/passwd"
(not very harmful/useful in this case, but if it had been a Tcl script
things could have gone worse)

What is the expected output? What do you see instead?
The problem is that:
    a) Apache runs the program with argv[0] set something that prevents
Tcl's unix/tclUnixFile.c:TclpFindExecutable() from finding itself; and
    b) Tcl then adjusts 'argv0' to match argv[1] in tclMain.c
    c) Tclkit's kitInit.c finally kicks in and SetExecName() uses the
contents of 'argv0' to find itself, erroneously locating whatever was in
argv[1].

What version of the product are you using? On what operating system?


Please provide any additional information below.

This is the third of three problems I encountered building tcl/tk 8.6.2b with 
kitgen 273ad4c... on Windows 7, using Visual Studio 2010.

Code in each of the 'rules.vc' files attempts to determine the compiler version 
in use and set VCVER accordingly.  That version number is later used to name 
certain library files (e.g. mk4vc<n>.lib).

Several of the rules.vc files use an if/then/elseif chain to map the compiler's 
internal version number (VCVERSION) to VCVER, but at least one of those fails 
to map VC9 and above correctly and others fail to map VC10 correctly.

Others of the rules.vc files use an arithmetic expression to map VCVERSION to 
VCVER.  That expression works correctly for all VC versions through 2010 (the 
newest I had available) and would presumably work correctly for newer versions 
as well.

The following patch changes each of the rules.vc files to use the arithmetic 
expression method of mapping VCVERSION to VCVER.

Subject: [PATCH] fixes for building on Windows with VC10

---
 8.x/itcl/win/rules.vc          |   13 +++++--------
 8.x/mk/win/rules.vc            |   15 +++++++++------
 8.x/thread/win/vc/nmakehlp.exe |  Bin 65536 -> 65536 bytes
 8.x/thread/win/vc/rules.vc     |   13 +++++--------
 8.x/vqtcl/win/rules.vc         |   13 +++++--------
 Makefile.vc                    |   17 +++++++----------
 6 files changed, 31 insertions(+), 40 deletions(-)
 mode change 100644 => 100755 8.x/thread/win/vc/nmakehlp.exe

diff --git a/8.x/itcl/win/rules.vc b/8.x/itcl/win/rules.vc
index 4039582..506d990 100644
--- a/8.x/itcl/win/rules.vc
+++ b/8.x/itcl/win/rules.vc
@@ -192,14 +192,11 @@ VCVER=0
 !if ![echo VCVERSION=_MSC_VER > vercl.x] \
     && ![cl -nologo -TC -P vercl.x $(ERRNULL)]
 !include vercl.i
-!if $(VCVERSION) >= 1500
-VCVER=9
-!elseif $(VCVERSION) >= 1400
-VCVER=8
-!elseif $(VCVERSION) >= 1300
-VCVER=7
-!elseif $(VCVERSION) >= 1200
-VCVER=6
+!if ![echo VCVER= ^\> vercl.vc] \
+    && ![set /a $(VCVERSION) / 100 - 6 >> vercl.vc]
+!include vercl.vc
+!endif
+!if ![del $(ERRNUL) /q/f vercl.x vercl.i vercl.vc]
 !endif
 !endif

diff --git a/8.x/mk/win/rules.vc b/8.x/mk/win/rules.vc
index d5f839c..737695a 100644
--- a/8.x/mk/win/rules.vc
+++ b/8.x/mk/win/rules.vc
@@ -184,15 +184,18 @@ VCVER=0
 !if ![echo VCVERSION=_MSC_VER > vercl.x] \
     && ![cl -nologo -TC -P vercl.x $(ERRNULL)]
 !include vercl.i
+!if ![echo VCVER= ^\> vercl.vc] \
+    && ![set /a $(VCVERSION) / 100 - 6 >> vercl.vc]
+!include vercl.vc
+!endif
+!if ![del $(ERRNUL) /q/f vercl.x vercl.i vercl.vc]
+!endif
+!endif
+
+# Since MSVC8 we must deal with manifest resources.
 !if $(VCVERSION) >= 1400
-VCVER=8
 _VC_MANIFEST_EMBED_EXE=if exist [email protected] mt -nologo -manifest [email protected] -outputresource:$@;1
 _VC_MANIFEST_EMBED_DLL=if exist [email protected] mt -nologo -manifest [email protected] -outputresource:$@;2
-!elseif $(VCVERSION) >= 1300
-VCVER=7
-!elseif $(VCVERSION) >= 1200
-VCVER=6
-!endif
 !endif

 #----------------------------------------------------------
diff --git a/8.x/thread/win/vc/nmakehlp.exe b/8.x/thread/win/vc/nmakehlp.exe
old mode 100644
new mode 100755
diff --git a/8.x/thread/win/vc/rules.vc b/8.x/thread/win/vc/rules.vc
index b94575a..6630600 100644
--- a/8.x/thread/win/vc/rules.vc
+++ b/8.x/thread/win/vc/rules.vc
@@ -192,14 +192,11 @@ VCVER=0
 !if ![echo VCVERSION=_MSC_VER > vercl.x] \
     && ![cl -nologo -TC -P vercl.x $(ERRNULL)]
 !include vercl.i
-!if $(VCVERSION) >= 1500
-VCVER=9
-!elseif $(VCVERSION) >= 1400
-VCVER=8
-!elseif $(VCVERSION) >= 1300
-VCVER=7
-!elseif $(VCVERSION) >= 1200
-VCVER=6
+!if ![echo VCVER= ^\> vercl.vc] \
+    && ![set /a $(VCVERSION) / 100 - 6 >> vercl.vc]
+!include vercl.vc
+!endif
+!if ![del $(ERRNUL) /q/f vercl.x vercl.i vercl.vc]
 !endif
 !endif

diff --git a/8.x/vqtcl/win/rules.vc b/8.x/vqtcl/win/rules.vc
index 0b78bc8..345bb6f 100644
--- a/8.x/vqtcl/win/rules.vc
+++ b/8.x/vqtcl/win/rules.vc
@@ -192,14 +192,11 @@ VCVER=0
 !if ![echo VCVERSION=_MSC_VER > vercl.x] \
     && ![cl -nologo -TC -P vercl.x $(ERRNULL)]
 !include vercl.i
-!if $(VCVERSION) >= 1500
-VCVER=9
-!elseif $(VCVERSION) >= 1400
-VCVER=8
-!elseif $(VCVERSION) >= 1300
-VCVER=7
-!elseif $(VCVERSION) >= 1200
-VCVER=6
+!if ![echo VCVER= ^\> vercl.vc] \
+    && ![set /a $(VCVERSION) / 100 - 6 >> vercl.vc]
+!include vercl.vc
+!endif
+!if ![del $(ERRNUL) /q/f vercl.x vercl.i vercl.vc]
 !endif
 !endif

diff --git a/Makefile.vc b/Makefile.vc
index c4d06ab..4442861 100644
--- a/Makefile.vc
+++ b/Makefile.vc
@@ -137,18 +137,15 @@ VCVER=0
 !if ![echo VCVERSION=_MSC_VER > vercl.x] \
     && ![cl -nologo -TC -P vercl.x >NUL]
 !include vercl.i
-!if ![del /q vercl.i vercl.x >NUL]
+!if ![echo VCVER= ^\> vercl.vc] \
+    && ![set /a $(VCVERSION) / 100 - 6 >> vercl.vc]
+!include vercl.vc
 !endif
-!if $(VCVERSION) >= 1500
-VCVER=9
-!elseif $(VCVERSION) >= 1400
-VCVER=8
-!elseif $(VCVERSION) >= 1300
-VCVER=7
-!elseif $(VCVERSION) >= 1200
-VCVER=6
+!if ![del $(ERRNUL) /q/f vercl.x vercl.i vercl.vc]
 !endif
 !endif
+
+# Since MSVC8 we must deal with manifest resources.
 !if $(VCVERSION) >= 1400
 _VC_MANIFEST_EMBED_EXE=if exist [email protected] mt -nologo -manifest [email protected] -outputresource:$@;1
 _VC_MANIFEST_EMBED_DLL=if exist [email protected] mt -nologo -manifest [email protected] -outputresource:$@;2
@@ -375,7 +372,7 @@ $(BUILD)\tclkit.res: ..\..\tclkit.rc
 .PHONY: setup files tidy clean

 files:
-   @$(CPY) ..\..\files $(BUILD)\files^\
+   @$(CPY) ..\..\files $(BUILD)\files\

 setup:
    @if not exist $(BUILD) mkdir $(BUILD)
-- 
1.7.9


-Steve Tarr

Isn't vlerq supposed to replace Mk4tcl? If yes - sdx doesn't always cover 
absence of Mk4tcl.

What steps will reproduce the problem?
1. recompile tclkit without Mk4tcl inside (vlerq is inside)
2. run cmds like sdx wrap, sdx mkpack...

What is the expected output? What do you see instead?
An error "invalid command name mk::file" occurs.

What version of the product are you using? On what operating system?
Any OS, latest sdx, recompiled tclkit without Mk4tcl inside.

Please provide any additional information below.
F.i. when executing "sdx wrap" or "sdx mkinfo", sdx requires presence any of 
vlerq or Mk4tcl package and uses the existent one. But, for operations like 
"sdx mkpack", "sdx mkshow", "sdx version" etc. the presence of Mk4tcl is 
strongly required.

I don't have UPX installed, so I had to modify makefile.include to not use it.  
Otherwise, the kitsh -init- step never gets done.

8.6b1 was released in 2008. Any chance we'd see a kit for that? 

coroutines would solve a lot of my problems...

This is the first of three problems I encountered building tcl/tk 8.6.2b with 
kitgen 273ad4c... on Windows 7, using Visual Studio 2010.

The build fails because Makefile.vc is copying some of the files to 
$(BUILD)/files^ rather than to $(BUILD)/files, where following steps expect to 
find them.  Am I missing something, or is this a simple typo in the Makefile.

The following patch fixes the problem, assuming it was just a typo.

Subject: [PATCH] fix typo in path

---
 Makefile.vc |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/Makefile.vc b/Makefile.vc
index c4d06ab..9e5dcd4 100644
--- a/Makefile.vc
+++ b/Makefile.vc
@@ -375,7 +375,7 @@ $(BUILD)\tclkit.res: ..\..\tclkit.rc
 .PHONY: setup files tidy clean

 files:
-   @$(CPY) ..\..\files $(BUILD)\files^\
+   @$(CPY) ..\..\files $(BUILD)\files\

 setup:
    @if not exist $(BUILD) mkdir $(BUILD)
-- 
1.7.9


-Steve Tarr

What steps will reproduce the problem?
1. Try to use the downloaded tclkit in a newer linux distro, eg in Karmic
Koala aka Ubuntu 09.10

What is the expected output? What do you see instead?
Expected: it runs
Instead: complains about missing libstdc++5

What version of the product are you using? On what operating system?
Using tclkit-8.5.7-linux-ix86 on Ubuntu-Karmic

Please provide any additional information below.
Newer distros (or at least ubuntu) stopped providing libstdc++5, it is not
even in the repos. This may be THEIR bug and not ours, but it will probably
turn out to be OUR problem not theirs :(

USER SOLUTION:
Just creating a symlink from libstdc++5 to libstdc++6 does not work, that
would have been too easy :(
What can be done is downloading the libstdc++5 package from the Jaunty
repos and manually installing it with dpkg, or else adding the Jaunty repos
and installing the package using synaptic or apt-get.

This will probably be a HUGE problem for many users.

What steps will reproduce the problem?
1. copy your starpack to a (remote) network drive
2. start it from there, measure the time
3. start it from a local copy (c:\...), compare the times

The times should be roughly comparable, but the network start is by ways slower.
Examining what's going on with procmon (sysinternals) shows that (of course) 
the starpack opens it's own EXE file to read it's VFS contents, and that seems 
to happen in very small chunks (512 byte or so).
The whole process is very slow.

See http://wiki.tcl.tk/26663 for a short discussion.

Tested many versions (8.4.x, 8.5.x), doesn't seems to matter.
Sure, I speak of MS WINDOWS (XP, Server 2003).

I don't have a trace handy, but if this is wanted, I can give it later.

What steps will reproduce the problem?
1. Obtain kitgen, either file or git
2. Read the included README file
3. Follow the step: make tars

What is the expected output? What do you see instead?

Expected: the files vfs.tar.gz, zlib.tar.gz, and vqtcl.tgz to be downloaded
and de'tar'd.
Instead: nothing is downloaded because TAR_URL is not defined

What version of the product are you using? On what operating system?

Both kitgen-master.tar.gz and kitgen via git are the same.
Solaris 10 (uname -a: SunOS x 5.10 Generic_142900-03 sun4u sparc SUNW,A70)

Please provide any additional information below.

Tried 'make configs' but that also failed on vqtcl using "Subversion
command-line client, version 1.4.6."

% svn checkout http://tclkit.googlecode.com/svn/trunk/vqtcl vqtcl
svn: REPORT request failed on '/svn/!svn/vcc/default'
svn: REPORT of '/svn/!svn/vcc/default': 400 Bad Request
(http://tclkit.googlecode.com)

I finally obtained vqtcl with a recursive wget on the svn trunk.

mk4vrs.tcl contains the following code:

    set mkfile [file join [file dirname $mkfile] \
                          [file readlink $mkfile]]

but [file readlink] can return an absolute path, resulting in an invalid 
$mkfile.  Suggest instead:

    set mkfile [file normalize [file join $mkfile __dummy]]

What steps will reproduce the problem?
1. Download tclkit windows 8.5.8 ix86 or amd64
2. Use Winzip 7 to create a 6gig zip file using "maximum compatibility" 
compression.  Note, the .zip is 6gig, individual files inside are on the order 
of a few megs to a few hundred megs.
3. Attempt to mount the 6gig zip using ::vfs::zip::Mount.  Both ix86 and amd64 
tclkit exe's return an error in zip::open on a seek call stating invalid 
parameter.  It appears to be a 32bit file offset limit issue.

What is the expected output? 

The .zip file should mount as a new vfs directory tree.

What do you see instead?

Error message from tclkit runtime.

What version of the product are you using? 

tclkit 8.5.8, tested with ix86 and amd64 versions.

On what operating system?

Win 7 64bit

Please provide any additional information below.

There has been a lot of fixes/improvements for zipvfs:
http://tclvfs.cvs.sourceforge.net/viewvc/tclvfs/tclvfs/library/zipvfs.tcl?view=l
og

some of them are crucial and should be incorporated into Tclkit. Basically the 
whole operation boils down to copying new zipvfs.tcl file into Tclkit's VFS, so 
it should be easy to build and deliver new version.

In the definition of _mount within mk4vfs.tcl one finds this sequence of code:

            set mode 0
            foreach arg $args {
                switch -- $arg {
                    -readonly   { set mode 1 }
                    -nocommit   { set mode 2 }
                }
            }
            if {$mode == 0} {
                periodicCommit $db
            }
            set v::mode($db) [lindex {translucent readwrite readwrite} $mode]

It would appear that the static list passed to lindex { } $mode as the last 
line should read:

{translucent readonly readwrite}

What version of the product are you using? 

tclkit-8.5.8

On what operating system?

Both Linux and Windows.

It was a big surprise that my app working with TclKit 8.5.2 is no more
working on TclKit 8.6 (b1).
the following script

 #----------------------------------------
  # load http (or any other package) in main thread
 package require http
  # OK

  # now create a thread and load http in (or any other package)
 package require Thread

 set t0 [thread::create]
 thread::send $t0 {
   package require http
 }
   # OK in 8.5.2
   # KO in 8.6b1.
   # KO in 8.5.6
   #    Error is "can't find package http"
 #----------------------------------------

I noticed that in 8.6b1 thread "auto_path" is not set !
( in 8.5.2 it is correctly set)

Moreover,
in 8.6b1 running
   thread::send $t0 "namespace children ::"
returns
  ::thread ::tsv ::tpool ::oo ::tcl
whilst in 8.5.2 it returns:
  ::thread ::mk ::tsv ::tpool ::mk4vfs ::vfs ::tcl

I also tried to set auto_path in the t0 thread

 thread::send $t0 [list set ::auto_path $::auto_path]
 thread::send $t0 {
   package require http
 }
but result (in 8.6) is always "can't find package http"

** Is it a bug in 8.6 or is there something wrong with my code ?
**  In case of bug, is there a workaround ?

thanks for your help 

** Note:
Currently the official supported version on this site is 8.5.2 (and it works).
I downloaded 8.5.6 and 8.6b1 from Pat Thoyts's site.

What steps will reproduce the problem?
1. run tclkit-8.5 sdx.kit md5sum <file>
2. <file> size should rather big, let's say 5Mb.
3. wait (almost) forever :)

What is the expected output? What do you see instead?
hex result from md5sum command expected.
But, this operation can lasts very long, almost forever.

What version of the product are you using? On what operating system?
win7 32b; tclkit-8.5.9; sdx.kit - any, even the latest.

Please provide any additional information below.

If run sdx.kit md5sum with tclkit-8.4 the operation is ok.

I found recently this problem and the solution I made: use new md5x.tcl 2.0.7 
from the latest tcllib 1.13 instead of this old md5.tcl 1.4.2. 

I know this has been an ongoing issue, I finally rolled up my sleeves
and dug into the sdx code. I figured out what the underlying problem
and solution is for replacing custom icons using tclkit / sdx to wrap
starkits. Pat Thoyts, please consider putting this fix in the sdx code
as soon as you can.

The problem:
when replacing the tcl icon with a custom icon in a starpack using the
latest tclkit (8.5.8), you will likely see the following messages:

  icon 48x48/256: NOT SAME SIZE
  icon 32x32/256: NOT SAME SIZE
  icon 16x16/256: NOT SAME SIZE
  icon 48x48/256: NOT SAME SIZE
  icon 32x32/256: NOT SAME SIZE
  icon 16x16/256: NOT SAME SIZE
  icon 48x48/256: replaced
  icon 32x32/256: replaced
  icon 16x16/256: replaced

This causes the icon in your starkit exe to only be partially replaced
(3 of the 9 images). I got to thinking, why does the 256 keep
repeating.

The explanation:

I unwraped the sdx kit and took a look. It turns out that in the sdx
code, the 256 represents the number of colors. However while opening
the ico file and inspecting the contents, if the sdx wrapping code
sees 0 it sets the number of colors to search for 256. However,
according to what I found on wikipedia 0 represents true color, which
of course Vista / 7 (and I presume at least XP / 2003) support. I
suppose this was a holdover from when the starpack solution was first
created.

The starpack code create an array of images found in both the original
icon and the new icon with the array element name of width X height X
numberofcolors. It then tries to match the array in the original with
the same one in the new icon file. If it finds one it checks the size.
Since the number of colors is incorrectly set to 256 for those above 8
bits, the sizes will not match.

The solution:

1. unwrap sdx.kit
2. navigate to the wrap.tcl file
3. in the procedure decICO, change the line
    lappend result ${w}x${h}/$cc $image
to
    lappend result ${w}x${h}/$bc $image

(bc is the bits per pixel ...
http://en.wikipedia.org/wiki/ICO_%28file_format%29
). We'll use the bits per pixel as part of the array index as opposed
to the number of colors.

4. rewrap sdx.kit
5. rewrap your application.

You will see the following output when you rewrap your app (assuming
you have the correct images in your ico file:

 customizing tclkit.ico in executable
  icon 64x64/32: replaced
  icon 48x48/32: replaced
  icon 32x32/32: replaced
  icon 24x24/32: replaced
  icon 16x16/32: replaced
  icon 48x48/24: replaced
  icon 32x32/24: replaced
  icon 24x24/24: replaced
  icon 16x16/24: replaced
  icon 48x48/8: replaced
  icon 32x32/8: replaced
  icon 24x24/8: replaced
  icon 16x16/8: replaced

I hope this ends this longstanding issue once and for all. Pat, please
let me know what you think.

Thanks,
Patrick 

1. put generated (or downloaded from here) tclkit (Linux) in path with spaces, 
let's say "/home/user/test space"
2. Run tclkit from there
3. Type "package require Tk"

An error will occur:
couldn't load file "{/home/user/test 
space/tclkit-8.5.9-linux-ix86/lib/tk8.5}/../libtk8.5.so": {/home/user/test 
space/tclkit-8.5.9-linux-ix86/lib/tk8.5}/../libtk8.5.so: cannot open shared 
object file: No such file or directory

Impact: impossible to execute such a starkit from with Tk from path with spaces.

OS: Linux, tclkit: 8.5.9 (downloaded), 8.5.10 (generated)

--------
file setupvfs.tcl has a bug:

diff -r1.1 -r1.2
481c481
<       \[list load \[file join \[list \$dir\] ..\

---
>       \[list load \[file join \$dir ..\

What steps will reproduce the problem?
1.
2.
3.

What is the expected output? What do you see instead?


What version of the product are you using? On what operating system?


Please provide any additional information below.

Here's a sample makefile:
all:
    @echo "Nothing to do."

hb.png: hb.xcf
    xcf2png hb.xcf

hb-%x8.png: hb.png
    @echo convert "$^" -colors 255 `echo "$@" | sed
's@^.*-\([0-9]*x[0-9]*\)x\([0-9]\).png@ -resize \1 -depth \2@'`  $@; \
    convert "$^" -colors 255 `echo "$@" | sed
's@^.*-\([0-9]*x[0-9]*\)x\([0-9]\).png@ -resize \1 -depth \2@'`  $@

hb-%x4.png: hb.png
    @echo convert "$^" -colors 15 `echo "$@" | sed
's@^.*-\([0-9]*x[0-9]*\)x\([0-9]\).png@ -resize \1 -depth \2@'`  $@; \
    convert "$^" -colors 15 `echo "$@" | sed
's@^.*-\([0-9]*x[0-9]*\)x\([0-9]\).png@ -resize \1 -depth \2@'`  $@

%.png.pgm: %.png
    pngtopnm -alpha "$^" > "$@"

%x4.png.ppm: %x4.png
    pngtopnm "$^" | ppmquant 15 > "$@"

%x8.png.ppm: %x8.png
    pngtopnm "$^" | ppmquant 255 > "$@"

%.png.ico: %.png.ppm %.png.pgm
    ppmtowinicon -andpgms -output "$@" $^

hb.ico: hb-16x16x4.png.ppm hb-16x16x4.png.pgm hb-16x16x8.png.ppm
hb-16x16x8.png.pgm hb-32x32x4.png.ppm hb-32x32x4.png.pgm hb-32x32x8.png.ppm
hb-32x32x8.png.pgm hb-48x48x4.png.ppm hb-48x48x4.png.pgm hb-48x48x8.png.ppm
hb-48x48x8.png.pgm
    ppmtowinicon -andpgms -output "$@" $^

clean:
    rm -f *.png.ico *.png.ppm *.png.pgm hb.ico hb-*.png hb.png

.PHONY: clean all

What steps will reproduce the problem?
1. Download from
http://code.google.com/p/tclkit/downloads/detail?name=tclkitsh-8.5.8-win32.upx.e
xe

What is the expected output? What do you see instead?
File should be saved to disk.  Instead, AVIRA claims it found an unwanted
program named "HIDDENEXT/Crypted" and quarantines it.  


What version of the product are you using? On what operating system?
AVIRA AntiVir version 10 running under Windows XP


Please provide any additional information below.
AVIRA describes the "detection" as type "HEUR/Crypted E".  Unfortunately,
their web site provides no additional information or descriptions.  

This also occurs with versions 8.5.7, 8.5.6, 8.4.17, and 8.4.11

I got kitgen (from patthoyts-kitgen-0602b4f.tar.gz) and tried to build a 
Solaris 10 Sparc executable.  It segmentation faults running tclkit-cli.  I 
have tried each Tcl/Tk version from 8.5.7 through 8.5.10.  I have tried gcc3 
and gcc4 and Solaris Studio 12.2.  They all core dump at the same step: 
"./tclkit-cli ../../validate.tcl".  My config.sh command was "sh config.sh 
thread mk cli".  I tried dropping "cli" and it still built it and core dumped.  
Is it possible to skip the vlerq version?

I tried building it on a similar Solaris 10 x86_64, no problems with gcc4.

Is there a version of kitgen known to work for Sparc?  Is there something I 
need to do differently for Sparc?

BTW: I had to patch vlerq.c because it was illegal C syntax according to 
Solaris cc:
diff -ru ../patthoyts-kitgen-0602b4f/./8.x/vqtcl/generic/vlerq.c 
./8.x/vqtcl/generic/vlerq.c
--- ../patthoyts-kitgen-0602b4f/./8.x/vqtcl/generic/vlerq.c     Tue Jun 14 
09:22:43 2011
+++ ./8.x/vqtcl/generic/vlerq.c Wed Sep 28 17:28:21 2011
@@ -40,7 +40,7 @@
 #endif

 #if defined(WORDS_BIGENDIAN)
-#if !defined(_BIG_ENDIAN) || _BIG_ENDIAN == ""
+#if !defined(_BIG_ENDIAN)
 #define _BIG_ENDIAN 1
 #endif
 #endif

This is the second of three problems I encountered building tcl/tk 8.6.2b with 
kitgen 273ad4c... on Windows 7, using Visual Studio 2010.

The build fails because 8.x/thread/win/vc/nmakehlp.exe is stored in the git 
archive with the wrong permissions.

The following patch fixes the permissions problem.

Subject: [PATCH] fix file permission problem

---
 8.x/thread/win/vc/nmakehlp.exe |  Bin 65536 -> 65536 bytes
 1 files changed, 0 insertions(+), 0 deletions(-)
 mode change 100644 => 100755 8.x/thread/win/vc/nmakehlp.exe

diff --git a/8.x/thread/win/vc/nmakehlp.exe b/8.x/thread/win/vc/nmakehlp.exe
old mode 100644
new mode 100755
-- 
1.7.9


-Steve Tarr

Sorry. I could not find a place to post questions. I am not sure if this is the 
right place.

I am trying to build a threaded executable using tclkit but I am not able to 
compile or find a threaded tclkit for linux-x86_64.

When I try to use kitgen it fails with the errors below.

Is everyone having this problem and that is why the tclkit at 
http://code.google.com/p/tclkit/ are non-threaded?

Any pointers on how to solve this or any other tool that I should use to do 
this?

I was able to generate a non-threaded one but I can not generate a threaded 
version. 

I followed the steps in the page: https://github.com/patthoyts/kitgen

I ran config.sh on the kitgen directory as:
sh config.sh 8.5/linux-x86_64 thread mk b64 dyn

I have tried the following TCL/TK packages with the same results:

tk8.5.9-src.tar.gz
tcl8.5.9-src.tar.gz

tk8.5.8-src.tar.gz
tcl8.5.8-src.tar.gz

tcl8.5.2-src.tar.gz
tk8.5.2-src.tar.gz

I am getting the errors below. They all seems to be in the 8.x directory.

linux-x86_64 $ make > make.log
../../../../8.x/thread/generic/threadPoolCmd.c: In function `TpoolWaitObjCmd':
../../../../8.x/thread/generic/threadPoolCmd.c:533: warning: cast to pointer 
from integer of different size
../../../../8.x/thread/generic/threadPoolCmd.c: In function `TpoolGetObjCmd':
../../../../8.x/thread/generic/threadPoolCmd.c:743: warning: cast to pointer 
from integer of different size
../../../../8.x/thread/generic/threadPoolCmd.c: In function `TpoolWorker':
../../../../8.x/thread/generic/threadPoolCmd.c:1223: warning: cast to pointer 
from integer of different size
../../../../8.x/vqtcl/generic/vlerq.c: In function `CoerceColumn':
../../../../8.x/vqtcl/generic/vlerq.c:1273: warning: 'seq' might be used 
uninitialized in this function
../../kitInit.c: In function `TclKitPathObjCmd':
../../kitInit.c:337: warning: assignment discards qualifiers from pointer 
target type
make: ./kit-cli: Command not found
make: *** [tclkit-dyn] Error 127

Thanks in advancce.