Registry warning?

Registry warning?

[Dave Goldberg]

Over the past couple weeks I've migrated my set up from XEmacs 21.4 to Emacs 23.1.1 (as provided through the Ubuntu repository).  Everything seems to work just fine, but I get this warning every time I start gnus:

Warning: The Gnus registry could not be loaded from ~/.gnus.registry.eioio, creating a new one; The Gnus registry could not be loaded from ~/.gnus.registry.eioio, creating a new one

~/.gnus/registry.eioio seems to be persistent - that is the file doesn't get removed when I exit so I don't understand what might be wrong with it when it wants to reload.  I don't use the registry very heavily - mostly just for doing split-fancy-with-parent and sure enough that works during the course of a session, but not after quiting and restarting.

I think I may have had the issue with XEmacs as well but didn't really notice it because the warning wasn't quite so prominent.

I don't know how to debug it - debug-on-error doesn't do anything since it's just a warning and doesn't get triggered.

Any pointers appreciated.

Thanks,

-- 
Dave Goldberg
david.goldberg6@verizon.net

Re: Registry warning?

[David Engster]

Dave Goldberg writes:
> ~/.gnus/registry.eioio seems to be persistent - that is the file
> doesn't get removed when I exit so I don't understand what might be
> wrong with it when it wants to reload.

The persistence of the registry depends on (featurep
'hashtable-print-readable), which is available on FSF Emacs >=23.2. Gnus
should probably say something more meaningful on older Emacsen...

-David

Re: Registry warning?

[Dave Goldberg]

> Dave Goldberg writes:
>> ~/.gnus/registry.eioio seems to be persistent - that is the file
>> doesn't get removed when I exit so I don't understand what might be
>> wrong with it when it wants to reload.

> The persistence of the registry depends on (featurep
> 'hashtable-print-readable), which is available on FSF Emacs >=23.2. Gnus
> should probably say something more meaningful on older Emacsen...

Thanks.

-- 
Dave Goldberg
david.goldberg6@verizon.net

Re: Registry warning?

[Ted Zlatanov]

On Sun, 04 Sep 2011 23:01:43 +0200 David Engster <deng@randomsample.de> wrote: 

DE> Dave Goldberg writes:
>> ~/.gnus/registry.eioio seems to be persistent - that is the file
>> doesn't get removed when I exit so I don't understand what might be
>> wrong with it when it wants to reload.

DE> The persistence of the registry depends on (featurep
DE> 'hashtable-print-readable), which is available on FSF Emacs >=23.2. Gnus
DE> should probably say something more meaningful on older Emacsen...

I thought EIEIO had its own persistence mechanism, not using
`hashtable-print-readable'?  That was one of the reasons I switched to
it, so serialization would Just Work.

Ted

Re: Registry warning?

[David Engster]

Ted Zlatanov writes:
> On Sun, 04 Sep 2011 23:01:43 +0200 David Engster <deng@randomsample.de> wrote: 
>
> DE> Dave Goldberg writes:
>>> ~/.gnus/registry.eioio seems to be persistent - that is the file
>>> doesn't get removed when I exit so I don't understand what might be
>>> wrong with it when it wants to reload.
>
> DE> The persistence of the registry depends on (featurep
> DE> 'hashtable-print-readable), which is available on FSF Emacs >=23.2. Gnus
> DE> should probably say something more meaningful on older Emacsen...
>
> I thought EIEIO had its own persistence mechanism, not using
> `hashtable-print-readable'?  That was one of the reasons I switched to
> it, so serialization would Just Work.

No. If you look in your mails somewhere around May, you'll see I wrote
this to you when discussing ERT tests on Emacs 22. EIEIO cannot
magically save any LISP structure (what about recursive structures, for
example?). It uses prin1 and hopes for the best, but you can provide
your own printer in the ':printer' slot.

-David

Re: Registry warning?

[Ted Zlatanov]

On Wed, 28 Sep 2011 09:29:33 +0200 David Engster <deng@randomsample.de> wrote: 

DE> Ted Zlatanov writes:
>> On Sun, 04 Sep 2011 23:01:43 +0200 David Engster <deng@randomsample.de> wrote: 
>> 
DE> Dave Goldberg writes:
>>>> ~/.gnus/registry.eioio seems to be persistent - that is the file
>>>> doesn't get removed when I exit so I don't understand what might be
>>>> wrong with it when it wants to reload.
>> 
DE> The persistence of the registry depends on (featurep
DE> 'hashtable-print-readable), which is available on FSF Emacs >=23.2. Gnus
DE> should probably say something more meaningful on older Emacsen...
>> 
>> I thought EIEIO had its own persistence mechanism, not using
>> `hashtable-print-readable'?  That was one of the reasons I switched to
>> it, so serialization would Just Work.

DE> No. If you look in your mails somewhere around May, you'll see I wrote
DE> this to you when discussing ERT tests on Emacs 22. EIEIO cannot
DE> magically save any LISP structure (what about recursive structures, for
DE> example?). It uses prin1 and hopes for the best, but you can provide
DE> your own printer in the ':printer' slot.

Thanks for repeating the information, I either forgot or misunderstood
it.  So can we provide a custom printer for Emacs 23.1 and older?

Ted

Re: Registry warning?

[David Engster]

Ted Zlatanov writes:
> On Wed, 28 Sep 2011 09:29:33 +0200 David Engster <deng@randomsample.de> wrote: 
>
> DE> Ted Zlatanov writes:
>>> On Sun, 04 Sep 2011 23:01:43 +0200 David Engster <deng@randomsample.de> wrote: 
>>> 
> DE> Dave Goldberg writes:
>>>>> ~/.gnus/registry.eioio seems to be persistent - that is the file
>>>>> doesn't get removed when I exit so I don't understand what might be
>>>>> wrong with it when it wants to reload.
>>> 
> DE> The persistence of the registry depends on (featurep
> DE> 'hashtable-print-readable), which is available on FSF Emacs >=23.2. Gnus
> DE> should probably say something more meaningful on older Emacsen...
>>> 
>>> I thought EIEIO had its own persistence mechanism, not using
>>> `hashtable-print-readable'?  That was one of the reasons I switched to
>>> it, so serialization would Just Work.
>
> DE> No. If you look in your mails somewhere around May, you'll see I wrote
> DE> this to you when discussing ERT tests on Emacs 22. EIEIO cannot
> DE> magically save any LISP structure (what about recursive structures, for
> DE> example?). It uses prin1 and hopes for the best, but you can provide
> DE> your own printer in the ':printer' slot.
>
> Thanks for repeating the information, I either forgot or misunderstood
> it.  So can we provide a custom printer for Emacs 23.1 and older?

Yes, for some definition of "we" (read: not me). ;-)

-David

Re: Registry warning?

[Ted Zlatanov]

On Thu, 29 Sep 2011 08:02:01 +0200 David Engster <deng@randomsample.de> wrote: 

DE> Ted Zlatanov writes:
>> On Wed, 28 Sep 2011 09:29:33 +0200 David Engster <deng@randomsample.de> wrote: 

DE> No. If you look in your mails somewhere around May, you'll see I wrote
DE> this to you when discussing ERT tests on Emacs 22. EIEIO cannot
DE> magically save any LISP structure (what about recursive structures, for
DE> example?). It uses prin1 and hopes for the best, but you can provide
DE> your own printer in the ':printer' slot.
>> 
>> Thanks for repeating the information, I either forgot or misunderstood
>> it.  So can we provide a custom printer for Emacs 23.1 and older?

DE> Yes, for some definition of "we" (read: not me). ;-)

Is there an API for this printer?  What functions must it provide?  We
can just reuse the old registry save/load code.

Thanks
Ted

Re: Registry warning?

[David Engster]

Ted Zlatanov writes:
> On Thu, 29 Sep 2011 08:02:01 +0200 David Engster <deng@randomsample.de> wrote: 
>
> DE> Ted Zlatanov writes:
>>> On Wed, 28 Sep 2011 09:29:33 +0200 David Engster <deng@randomsample.de> wrote: 
>
> DE> No. If you look in your mails somewhere around May, you'll see I wrote
> DE> this to you when discussing ERT tests on Emacs 22. EIEIO cannot
> DE> magically save any LISP structure (what about recursive structures, for
> DE> example?). It uses prin1 and hopes for the best, but you can provide
> DE> your own printer in the ':printer' slot.
>>> 
>>> Thanks for repeating the information, I either forgot or misunderstood
>>> it.  So can we provide a custom printer for Emacs 23.1 and older?
>
> DE> Yes, for some definition of "we" (read: not me). ;-)
>
> Is there an API for this printer?  What functions must it provide?  We
> can just reuse the old registry save/load code.

It must do what `prin1' does, minus the optional PRINTCHARFUN
argument. That means, it has to print the slot in a way that can later
be understood by `read'. See also section 3.2 of the EIEIO info:

(info "(eieio)Slot options")

-David