Re: [TCLCORE] Bytecode compiler and monotonic time

[Harald O. <har...@el...>]

Am 14.05.2025 um 07:03 schrieb Christian Werner:
> On 05/13/2025 10:18 PM, Harald Oehlmann wrote:
> 
> Howdy Harald, all,
> 
>> ...
>> I am in the monotonic after event issue and I am amazed by the 
>> solution now in this ticket: https://core.tcl-lang.org/tcl/ 
>> info/1e2c6ce4c8a883c9
>>
>> TIP 233 is aparrently in the way of such a solution. Are there any 
>> comments on this? Andreas gave insights within the Telco and asks 
>> Kevin for his opinion. The use-case may be outdated and the TIP may be 
>> eventually removed.
> 
> while we're at the longstanding monotony issue wouldn't this be a good 
> opportunity
> to repair the per slave interpreter time limits? Those are currently 
> expressed as
> an absolute point in time in the wall clock domain. My more natural 
> thinking is to
> give that interp a certain amount of fuel in form of an interval 
> expressed in the
> time domain which the after command uses, in contrast to say tomorrow at 
> midnight
> your tank is empty.
> 
> BR,
> Christian
Thanks, Christian,

Are you speaking about the modifications in generic/tclInterp.c in the 
bug branch ?
https://core.tcl-lang.org/tcl/timeline?r=tkt3328635-posix-monotonic-clock&c=2025-05-14+06%3A30%3A15

You can also see it in this diff:
https://core.tcl-lang.org/tcl/info/8ea9c4081c39b8bf

When I look to "interp limit" man page, there, the limit is documented 
as milliseconds and seconds, and not as a fix time.

So, this is a bug and not a design change?

I see the usage of the monotonic clock in the branch for the following 
commands:

- interpreter limit
- time
- after

For me, all those are documented in time difference (e.g. monotonic 
clock) and not in wall clock difference.
So, the delay/speedup by a wall clock change is IMHO a bug and not a 
documented feature.

My own use-case does not cover "time" and "interpreter limit", but I 
see, that this is a bug. I never use slave interpreters. The image photo 
command has issues with it (as reported by Emiliano) and this is another 
story to dig in...

Thanks,
Harald