[elephant-devel] add-to-root problem

[Ian Eslick]

Hi All,

I just integrated, tested and pushed my local patches to the main 091  
repository, including fixing all the bugs I found under Allegro due  
to missing test dependencies.  The data store and migration tests are  
green on:

Mac OS X 32-bit / SBCL / BDB
Mac OS X 32-bit / Allegro 8.0 / BDB
Mac OS X 32-bit / SBCL / BDB
Mac OS X 32-bit / Allegro 8.0 / SQLite3

Please run this over postmodern and on any other platforms to make  
sure I didn't miss anything.  Slobodan, would you run this on LW on  
Windows?

We should discuss features and the test coverage that we are  
targeting as part of 0.9.1 so we can wrap up the current branch.   
Most of my edits were bug fixes or in my contrib folder, so this is  
primarily a postmodern release plus a little of my in-progress work  
on the elephant core.

The big projects I have outstanding (and which are likely to remain  
so for some time) are:
- Lisp prevalence data store (first pass is done)
- Query subsystem (anyone want to take this on, I can advise?)

There are also a set of outstanding tickets in Trac.  We should  
figure out which should be fixed before we release.  I'll look into  
#50 as it seems serious.  I think we should address any outstanding  
sorting issues as well.

Regards,
Ian

PS - Note on 5AM conversion:

The original tests assumed textual order, FiveAM test execution under  
allegro was random - perhaps under SBCL it accidentally maintained -  
but this required fixing a bunch of dependencies.

There are still some funny things going on with class finalization  
that I have to admit I don't understand fully; but it seems to have  
to do with the way fiveam stores tests to be run later.  RT stored  
them as sexps to be evaluated later, fiveam stores them as closures.   
It may be that external variables are captured as values and not as  
dynamic references...

On Oct 19, 2007, at 12:28 PM, Ian Eslick wrote:

> Ok, I've got a merged version of my 091 tree and the main 091  
> tree.  I get two errors under SBCL, but about 30 under Allegro.   
> Some of the changes on the main branch have seriously broken  
> Allegro with BDB.
>
> I'm looking into this.  Some of these errors may very well be  
> related to the problems showing up in Lispworks on Win32 as well.
>
> Stay tuned.
>
> Ian
>
> On Oct 8, 2007, at 4:47 PM, Slobodan Blazeski wrote:
>
>> I forget to mention, this is compiled with Allegro 7. Is version 7  
>> still supported ?
>>
>> Slobodan
>>
>>
>> On 10/8/07, Robert L. Read <read at robertlread.net> wrote: Ian, you  
>> use Allegro, don't you?
>>
>> The failure below is the place to look.
>>
>> "is not yet finalized" sure sounds like a difference between lisps or
>> compiler switches in some way.
>>
>>
>> On Mon, 2007-10-08 at 21:52 +0200, Slobodan Blazeski wrote:
>> >
>> >  Failure Details:
>> >  --------------------------------
>> >  INDEXING-BASIC-TRIVIAL []:
>> >       Unexpected Error: #<PROGRAM-ERROR @ #x213cebf2>
>> > Class #<PERSISTENT-METACLASS IDX-ONE> is not yet finalized...
>>
>>
>>
>>
>> _______________________________________________
>> elephant-devel site list
>> elephant-devel at common-lisp.net
>> http://common-lisp.net/mailman/listinfo/elephant-devel
>
> _______________________________________________
> elephant-devel site list
> elephant-devel at common-lisp.net
> http://common-lisp.net/mailman/listinfo/elephant-devel