[Ecls-list] Question #2

[Gabriel Dos Reis]

On Mon, Feb 15, 2010 at 3:29 PM, Tobias C. Rittweiler <tcr at freebits.de> wrote:
> Gabriel Dos Reis <gdr at integrable-solutions.net>
> writes:
>
>> On Mon, Feb 15, 2010 at 2:38 PM, Tobias C. Rittweiler <tcr at freebits.de> wrote:
>>
>>> LOADing fasls created by earlier versions of ECL into a more current
>>> version results in _undefined behaviour_. It might work, or might not.
>>
>> indeed, that has been reported to me several time in the past
>> (in connection with OpenAxiom.)
>>
>>> in case it does not
>>> match, an error would be signalled (ideally one with two restarts: one
>>> which results in recompilation, the other which forces loading.)
>>
>> I welcome the idea.  For OpenAxiom (and I suspect many applications
>> delivered to end-users), the restart would be a nuisance though.
>
> How can restarts be nuisances?  You mean because users are dropped into
> the debugger?

yes; most OpenAxiom users don't really care that behind the scene we
are using a Lisp system.  And dropping a end user into a debugger, for
a delivered application, is not what I would consider a user-friendly interface.

> In the latter case, you can get rid of that with the newly
> introduced *INVOKE-DEBUGGER-HOOK*!

yeah, now I have to track which versions of ECL is being used.