Re: [TCLCORE] Tcl/Tk 8.6.10 Release Candidates

[Donald P. <d.g...@co...>]

> On Nov 21, 2019, at 6:57 AM, Jan Nijtmans <jan...@gm...> wrote:
> 
> Op do 21 nov. 2019 om 12:51 schreef Donald Porter:
>> No, the main difference is the earlier VC++ 6.0 build failure reported in 3.30.1 and fixed in
>> 
>> http://cyqlite.sourceforge.net/cgi-bin/sqlite/info/23d9040438dfe202
> 
> I think you're mistaken, that one is already fixed in 3.30.1, which
> should build fine unmodified.

Sorry, I cited the wrong checkin.

Sqlite-3.30.1 was released on October 11

https://sourceforge.net/p/tcl/mailman/message/36783217/ <https://sourceforge.net/p/tcl/mailman/message/36783217/>

It was also included in the RC1 for tcl8.6.10. A build failure report came in and you made the fix.

https://sourceforge.net/p/tcl/mailman/message/36802112/ <https://sourceforge.net/p/tcl/mailman/message/36802112/>

Fix checked in here:

http://cyqlite.sourceforge.net/cgi-bin/sqlite/info/3edbcd1d6c5722fd <http://cyqlite.sourceforge.net/cgi-bin/sqlite/info/3edbcd1d6c5722fd>

A checkin dated November 6 cannot be part of a release made on October 11, no matter whether you move the tag.

So to get a bundled release that might build, I tried to cut my own sqlite-3.30.1.1 to include that build fix. We now know that has also failed, probably because the build system was never designed to deal with a 4-part version number.

That’s where we are. If we’re willing to accept a broken VC++ 6.0 build, the best answer is to drop back to the failing 3.30.1. If not, I need some help making the build system able to tolerate a version number 3.30.1.1.

DGP