atlsh
=====

The ATL8.5.20 source distribution :

* https://github.com/aotto1968/atlsh

is the continuation of the :

* https://www.tcl-lang.org/software/tcltk/8.5.html - tcl8.5.19

development with the goal of offering a fast, simple and complete language that is based on TCL but still enables new innovative
concepts.

history
=======

* 23 apr 2025 - first public announcement.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 23.04.25 14:09, aotto1968 wrote:

atlsh
=====

The ATL8.5.20 source distribution :

  * https://github.com/aotto1968/atlsh

is the continuation of the :

  * https://www.tcl-lang.org/software/tcltk/8.5.html - tcl8.5.19

development with the goal of offering a fast, simple and complete language that is based on TCL but still enables new innovative
concepts.

history
=======

  * 23 apr 2025 - first public announcement.

The porting of the first tcl8.6.20 commands to ATL8.5.20 has been successfully completed.
The following commands have been ported: try/catch/finally, throw, lsearch, lsort, and lassign.

ATL8.5.20 inherits the "tests" directory from tcl8.6.20, ensuring that all "possible" tcl8.6.20 commands run identically under
ATL8.5.20. The goal is maximum compatibility with an estimated 20% speed advantage by eliminating the "dead end" of NRE and TCLOO.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hi,

I had the requirement to add a specific feature to my development editor "vim" to understand the
development tree and modify the behavior depending on the local directors:

the @grok AI and I develop together the following solution:

here is the conversation: https://x.com/i/grok?conversation=1915671316327256554

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

The new URL of "atlsh" is online

http://thedev.nhi1.de/theATL/main/index.htm

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

add a new tool

atltest is a tool for testing the *.test files from the theATL/tests directory.

* The test file (*.test) from the theATL/atlsh/tests directory is mirrored to the theATL/tests directory via a soft link.
** The theATL test environment is created without changes to the atlsh distribution.
* Massive parallel testing with the automake make check was implemented in the theATL/tests directory.
** An adaptation to the tcltest tool and the *.test files was required.
* The largest adaptation was made to the test cases (*.test) provided by TCL. ** By default, TCL only supports sequential testing.
** By default, TCL uses the same temporary file name across different test cases.

http://thedev.nhi1.de/theATL/main/index.htm http://thedev.nhi1.de/theATL/main/atltest.htm

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

UPDATE

[atlsh-1-2 c317774] strategy change from tcl8.5+PATCH to tcl8.6-PATCH
1019 files changed, 141526 insertions(+), 135312 deletions(-)

During testing of tcl8.5 with the tcl8.6 test cases, it became apparent that there are numerous
serious (core) bugs in tcl8.5 that were only patched in tcl8.6. However, tcl8.6 is now the problematic
release with NRE and TCLOO. In a first attempt to patch tcl8.5, a large amount of code had to be
imported from tcl8.6, which then raises the question of whether it even makes sense to patch tcl8.5
towards tcl8.6 if the end result is tcl8.6 anyway. The strategy of starting with tcl8.6 and removing
everything that is unwanted seems to make more sense to me because it starts with a bug-free launch
that is then successively retested at every step.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 02.05.25 12:03, aotto1968 wrote:

UPDATE

[atlsh-1-2 c317774] strategy change from tcl8.5+PATCH to tcl8.6-PATCH
 1019 files changed, 141526 insertions(+), 135312 deletions(-)

During testing of tcl8.5 with the tcl8.6 test cases, it became apparent that there are numerous
serious (core) bugs in tcl8.5 that were only patched in tcl8.6. However, tcl8.6 is now the problematic
release with NRE and TCLOO. In a first attempt to patch tcl8.5, a large amount of code had to be
imported from tcl8.6, which then raises the question of whether it even makes sense to patch tcl8.5
towards tcl8.6 if the end result is tcl8.6 anyway. The strategy of starting with tcl8.6 and removing
everything that is unwanted seems to make more sense to me because it starts with a bug-free launch
that is then successively retested at every step.

After switching from tcl85 to tcl86, the paralellization of "tcltest" is complete.
- "zlib" and "tcloo" have also been removed.
- Next, "NRE" and "thread" support will be deleted.
- Once all this is complete, the "interp" command will be deleted and the technology will be rebuilt.

example: new massive parallel testing of the tcl86 test suite.
...
PASS: dict.test
PASS: exec.test
PASS: expr.test
PASS: reg.test
PASS: socket.test
PASS: thread.test
PASS: interp.test
PASS: lmap.test
PASS: async.test
PASS: chanio.test
PASS: tcltest.test
PASS: main.test
PASS: clock.test
PASS: regexp.test
PASS: io.test
PASS: ioCmd.test ============================================================================ Testsuite summary for NHI1 10.0 ============================================================================
# TOTAL: 134
# PASS: 134
# SKIP: 0
# XFAIL: 0
# FAIL: 0
# XPASS: 0
# ERROR: 0 ============================================================================ ...

real 1m30,248s
user 3m55,929s
sys 0m15,400s

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

atlsh-1-0-2 is out

- http://thedev.nhi1.de/theATL/main/index.htm
- https://github.com/aotto1968/atlsh

update 8 mai 2025
=================

- After switching from @tcl85 to @tcl86, it became apparent that there was a problem distinguishing an @b existing
@tcl86 installation from the new @atl10 installation.
- To circumvent this and protect @tcl86, the distribution was renamed to @atl10, thus changing the @e visible interface
to @b atl.
- The invisible interface (i.e., functions, variables, constants, etc.) remains untouched and thus binary compatible.
- However, this will change in the future when \c atlsh becomes a standalone product.
- The following changes were made:
\verbatim
from | to
------------------ | ----------------
tclConfig.sh | atlConfig.sh
tclsh | atlsh
tcltest | atltest
include/tcl*.h | include/atl/tcl*.h
lib/tcl8 | lib/atl1
lib/tcl8.6 | lib/atl1.0
bin/tclsh | bin/atlsh
lib64/pkgconfig/tcl.pc | lib64/pkgconfig/atl.pc
lib64/libtclstub8.6.a | lib64/libatlstub1.0.a
lib64/tclConfig.sh | lib64/atlConfig.sh
lib64/libtcl8.6.so | lib64/libatl1.0.so
share/man1/*.1 | share/atl/man1/*.1
share/man3/*.3 | share/atl/man3/*.3
share/mann/*.n | share/atl/mann/*.n
TCLLIBPATH | ATL_LIBPATH
TCL_LIBRARY | ATL_LIBRARY
\endverbatim

update 5 mai 2025
=================

After switching from @tcl85 to @tcl86, the paralellization of @tcltest was complete.
- @zlib and @tclOO have also been removed.
- all \e sub-packages from \c pkgs were removed.
- Next, @NRE and @thread support will be removed.
- Once all this is complete, the @interp command will be replaced by a \b NEW technology.
- example: new massive parallel testing of the @tcl86 test suite.
\verbatim
...
PASS: dict.test
PASS: exec.test
PASS: expr.test
PASS: reg.test
PASS: socket.test
PASS: thread.test
PASS: interp.test
PASS: lmap.test
PASS: async.test
PASS: chanio.test
PASS: tcltest.test
PASS: main.test
PASS: clock.test
PASS: regexp.test
PASS: io.test
PASS: ioCmd.test
============================================================================
Testsuite summary for NHI1 10.0
============================================================================
...

real 1m30,248s
user 3m55,929s
sys 0m15,400s
\endverbatim

update 2 mai 2025
=================

- Testing @tcl85 with the @tcl86 test cases revealed that @tcl85 contains numerous serious (core) bugs that were only
fixed in @tcl86.
- However, @tcl86 is now the problematic version with @NRE and @tclOO.
- An initial attempt to patch @tcl85 required importing a large amount of code from @tcl86.
- This raises the question of whether it even makes sense to patch @tcl85 towards @tcl86 when the end result is @tcl86 anyway.
- The strategy of starting with @tcl86 and removing everything unwanted seems more sensible to me, as it starts with a
bug-free launch, which is then successively retested at each step.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Currently, the build system is being ported from "atlsh" (alias: tcl8.6) to automake/libtool - the goal is to simplify
maintenance and remove as much "ballast" as possible that has accumulated in tcl over the years.

https://www.facebook.com/permalink.php?story_fbid=pfbid05bGd69jcmgxJTTBiiwh9dgvEz4fDcxXMSEsqSxR8Pneyp93aCkms9qoSQWQs79EBl&id=100069563501101

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)