Version 1.12.1 of the Magicsplat Tcl/Tk 8.6.12 for Windows binary
distribution is released with new packages and updates.

Home page: https://www.magicsplat.com/tcl-installer/index.html
Changes: https://www.magicsplat.com/tcl-installer/index.html#changes

Ashok,

Thank you very much for your efforts on promoting Tcl/Tk for Windows!

I have some remarks, hopefully useful to you.

1) You build 64-bit version as well as 32-bit which is very very good because there are still a lot of 32-bit machines in use. But imagine there are so ancient versions of Windows 32-bit that have no MSI option to install .msi (only .exe), and being unsupported they cannot be upgraded for this option. I mean Windows XP. At that those platforms are still alive and in good health:) - would you consider to upload also .exe installer for Windows 32-bit?

2) There is some trouble with access to https://sourceforge.net from Russia at least, obviously due to those sanctions you know. But http://sourceforge.net is working, still no guarantees it will work tomorrow. Probably, Bill Gates is much more free and independent on the politicians than SF guys (though who knows?). Do you consider to mirror your Tcl/Tk distributions at GitHub too? They have the Releases option in GitHub, very convenient.

3) Also, I'm touched with your appreciation of my humble efforts in Tcl/Tk. But I should note that apave package contains baltip, bartabs and hl_tcl packages. So no need to include them separately. The apave package alone will be enough. Those interested will find the baltip, bartabs and hl_tcl packages in apave directory. And you save your efforts and the distribution's size a bit.

Regards,
Alex

On 6/13/2022 1:14 PM, Alex P wrote:
> Ashok,
>
> Thank you very much for your efforts on promoting Tcl/Tk for Windows!
>

yw

> I have some remarks, hopefully useful to you.
>
> 1) You build 64-bit version as well as 32-bit which is very very good because there are still a lot of 32-bit machines in use. But imagine there are so ancient versions of Windows 32-bit that have no MSI option to install .msi (only .exe), and being unsupported they cannot be upgraded for this option. I mean Windows XP. At that those platforms are still alive and in good health:) - would you consider to upload also .exe installer for Windows 32-bit?

The issue is more than the installer. The binaries are built with a
runtime that is not available on XP. Building with an XP supported
runtime had some issues that I don't recall (likely solvable but not
where I want to spend my time). Paul's BAWT distribution on the other
hand is (I think, not sure) built with gcc and therefore should fill the
need.

Moreover, as a rule, I'm reluctant to continue supporting what are now
ancient platforms. There is always a non-zero cost involved. If I claim
XP support, I have to test it.

>
> 2) There is some trouble with access to https://sourceforge.net from Russia at least, obviously due to those sanctions you know. But http://sourceforge.net is working, still no guarantees it will work tomorrow. Probably, Bill Gates is much more free and independent on the politicians than SF guys (though who knows?). Do you consider to mirror your Tcl/Tk distributions at GitHub too? They have the Releases option in GitHub, very convenient.
>

May be at some point. Right now my scripts are set up with SF as
distribution point for all my extensions. Not currently a priority to
update them. If in fact SF becomes unavailable, I'll see. Incidentally,
the distribution is also available via Chocolatey.

> 3) Also, I'm touched with your appreciation of my humble efforts in Tcl/Tk. But I should note that apave package contains baltip, bartabs and hl_tcl packages. So no need to include them separately. The apave package alone will be enough. Those interested will find the baltip, bartabs and hl_tcl packages in apave directory. And you save your efforts and the distribution's size a bit.
>

Ok, good to know.

/Ashok

Am 13.06.2022 um 04:48 schrieb apn:
> Version 1.12.1 of the Magicsplat Tcl/Tk 8.6.12 for Windows binary
> distribution is released with new packages and updates.
>
> Home page: https://www.magicsplat.com/tcl-installer/index.html
> Changes: https://www.magicsplat.com/tcl-installer/index.html#changes

Dear Ashok,
it would be great, if you could also update the TCL+TWAPI starkits.

That would help to try the new code related to my observed TLS bug.

Thank you and take care,
Harald

Similar to Ashok I do not support XP, but invested 2 hours to give it a try.
Using a XP virtual machine, which I did not power on since 10 years gave the following results:

The InnoSetup based BAWT installer does not work on XP.
If you extract the binaries (i.e. copy folder Tcl to ex. c:\Tcl) and add C:\Tcl to your Path variable,
most of the simple BAWT tests run fine with the exception of:
cffi
iocp
mawt
tclpy
tls
trf
twapi

Tests done with Tcl 8.6.12 32-bit.
Tcl 8.7 does not work at all.

Paul

Am 14.06.2022 um 07:37 schrieb apn:
> On 6/13/2022 1:14 PM, Alex P wrote:
>> Ashok,
>>
>> Thank you very much for your efforts on promoting Tcl/Tk for Windows!
>>
>
> yw
>
>> I have some remarks, hopefully useful to you.
>>
>> 1) You build 64-bit version as well as 32-bit which is very very good because there are still a lot of 32-bit machines in use. But imagine there are so ancient versions of Windows 32-bit that have no MSI option to install .msi (only .exe), and being unsupported they cannot be upgraded for this option. I mean Windows XP. At that those platforms are still alive and in good health:) - would you consider to upload also .exe installer for Windows 32-bit?
>
> The issue is more than the installer. The binaries are built with a runtime that is not available on XP. Building with an XP supported runtime had some issues that I don't recall (likely solvable but not where I want to spend my time). Paul's BAWT distribution on the other hand is (I think, not sure) built with gcc and therefore should fill the need.
>
> Moreover, as a rule, I'm reluctant to continue supporting what are now ancient platforms. There is always a non-zero cost involved. If I claim XP support, I have to test it.
>
>>
>> 2) There is some trouble with access to https://sourceforge.net from Russia at least, obviously due to those sanctions you know. But http://sourceforge.net is working, still no guarantees it will work tomorrow. Probably, Bill Gates is much more free and independent on the politicians than SF guys (though who knows?). Do you consider to mirror your Tcl/Tk distributions at GitHub too? They have the Releases option in GitHub, very convenient.
>>
>
> May be at some point. Right now my scripts are set up with SF as distribution point for all my extensions. Not currently a priority to update them. If in fact SF becomes unavailable, I'll see. Incidentally, the distribution is also available via Chocolatey.
>
>> 3) Also, I'm touched with your appreciation of my humble efforts in Tcl/Tk. But I should note that apave package contains baltip, bartabs and hl_tcl packages. So no need to include them separately. The apave package alone will be enough. Those interested will find the baltip, bartabs and hl_tcl packages in apave directory. And you save your efforts and the distribution's size a bit.
>>
>
> Ok, good to know.
>
> /Ashok
>

Thank you, Ashok and Paul, for your detailed answers and concern. It's a general problem with those obsolete technologies, when you should and still don't like to say bye to them. All those extensions being abandoned... SF being replaced by GH... I don't want to continue with the listing.

Paul, I'm touched with your appreciation of my apave too. By the way, I'd forked your poApps just to include in poDiff the feature of comparing directories by list, perhaps you remember my email about this. It's of a great help for me, because, indeed, I keep my packages in several locations that should be consistent and up to date. You can see the fork here:

https://github.com/aplsimple/poApps

The files 1.podiff and 2.podiff make the work. The implementation of this idea is in no way ideal, so I hope you would take it as a base and implement it as it should be. Thank you so much.

On 6/14/2022 2:43 PM, Harald Oehlmann wrote:
> Dear Ashok,
> it would be great, if you could also update the TCL+TWAPI starkits.
>
> That would help to try the new code related to my observed TLS bug.
>
> Thank you and take care,
> Harald

I've updated the tclkits.

Regarding the TLS, I don't think there will be a difference. Between
autoproxy, http and tls it is difficult to tell what's going on. It
seems to me there is a callback running after the socket is closed. This
should be protected against but I'm not clear on exactly where.

I shall take another look later.

/Ashok