Removing DirectoryResource does not remove its children. ?

Alexandre Rafalovitch writes:
 > Hi,
 > 
 > Still playing around with DirectoryResource and feel I found so many
 > bugs/inconsistancies, that maybe that is me, who does not understand the
 > arcitecture and trying to fit every possible thing where it does not
 > belong? :-{

Let's see...

 > The question is thus:
 > I have a DirectoryResource. I delete it, because it has many entries
 > pointing to non existant files. The reason why my DirectoryResource has
 > pointers to non-existing files is because it never deletes resources, when
 > their files are deleted. I could call it another bug, I guess.( You know
 > there is 'extending' flag, but there is no 'contracting' flag :-{} ).
 > Anyway, i delete a DirectoryResource with hopes that when I index it again,
 > it would be the same result, as if it never was indexed again. However,
 > when it is entered next time, I see old familiar directory listing with
 > many non existing entries. Tell me it is not confusing. The only way I see
 > now to do that is to go and select everything in the directory, mark it for
 > reindexing  or deleting and go. However, trying to select 50 or so
 > checkboxes can be frustrating..... :-{

Ok, the problem is the following. When you delete the directory
resource, what you do is you remove it from it's parent resource
store (i.e it's parent .jigidx file). However, as the sub-directory
has already been indexed, it has its own resource store (namely
foo/.jigidx), which contains the description of all the resource in
this sub-directory. When you re-create the directory resource, it
aumatically (!) thinks that the old .jigidx file  is for it.

In brief, erase the sub-directory .jigidx file in order to acheieve
your goal. This is again a configuration problem that should/will be
addressed in the next version (deleteing a directory resource should
delete its resource store).

The extending flag ought to also be a contracting flag, this again
will be fixed in next release (hopefully).
 
 > I start to think that this, and several other problems can be solved with
 > things discussed in previous mail such as 'Indexable' flag and recursive
 > operations, but such things have to be discussed in details, so all cases
 > are covered.

Yes, that's the way it is going to be, you are right.

Anselm.

Received on Thursday, 20 June 1996 14:08:59 UTC