Re: `vocabulary` broken?

On Sat, Sep 27, 2025, at 3:32 PM, Bernd Paysan wrote:
> Am Samstag, 27. September 2025, 13:56:10 Mitteleurop=C3=A4ische Sommer=
zeit schrieb=20
> Ethan Azariah:
>> I can't reproduce it on my 64-bit tablet either, and I regularly use =
lines
>> like this in 0.7.0. Of my various devices, the problem only occurs on=
 a
>> 32-bit x86 tablet, Android 7.1.2 (LineageOS 14.1), Gforth 0.7.9_20250=
305.
>> I'm very surprised to have got a 32-bit tablet, but now I have and the
>> older, rooted Android is good for tinkering.
>>=20
>> I remember from years ago some weird heisenbug affected only a 32-bit
>> Android device. Do we want to track it down or do you just want to ca=
ll
>> 32-bit obsolete? I don't mind; Gforth is convenient but a bit too com=
plex
>> for me to really get into. I could run debugging binaries images or
>> whatever if you want to supply them, but I don't want to use the Andr=
oid
>> SDK and I'm not yet used to using a debugger.
>
> It should be sufficient to look at what=20
>
> ORDER
>
> prints out.  It should be
>
> Forth Forth Root   Forth

I put in `order` as a debug print before the block was loaded. The bug a=
ppeared whether the first item was `editor` (default in my startup seque=
nce) or `Forth` (set by `only forth also`).

The bug has disappeared now I've moved the loading of the qxv block to b=
efore the editor. I can't see any reason why it should have disappeared,=
 nor why it appeared in the first place. This is why I think 32-bit Gfor=
th has, in recent years, got a bug not present in 64-bit. The last time =
I used Gforth on 32-bit Android, there was a similar bug with code not w=
orking which worked everywhere else or started to work when I changed so=
mething irrelevant. At the time, I didn't think it was worth fixing and =
I can't get details now because that device is broken.