[Zope-CMF] [dev] Document.get_size considered dumb

[Florent Guillaume]

In article <cis3r9$oba$1 at sea.gmane.org> you write:
> I would like to modify the implementation of 
> Products.CMFDefault.Document.get_size as follows:
> 
> 
> --- CMFDefault/Document.py      7 Sep 2004 08:47:54 -0000       1.69.2.2
> +++ CMFDefault/Document.py      22 Sep 2004 14:53:13 -0000
> @@ -415,6 +415,6 @@
>       security.declareProtected(View, 'get_size')
>       def get_size( self ):
>           """ Used for FTP and apparently the ZMI now too """
> -        return len(self.manage_FTPget())
> +        return len(self.text)
> 
>   InitializeClass(Document)
> 
> The current behavior tries to ensure that 'get_size' returns exactly the 
> same value as the length of the decorated source text.  The only 
> dependent I know of for that behavir is webdav's HEAD;  I have no 
> problem with "breaking" that, because there is no guarantee that HEAD 
> and GET will return the same values between two requests, anyway.
> 
> Are there any other clients out there who care about 'get_size'?  If 
> noone objects, I will check it in on Friday (I may make the size cached, 
> as well).

No pb with that.

As we've had requests from people about it, I'd like to confirm:
get_size() is only used for "informative human-readble" values of the
size, right ?

A use case we've had is for an objects that stores, for instance, a PDF
file, but also internal full-text versions for the indexer and other
things. The user expects to have get_size() return the size of the PDF,
but in some cases (quota accounting) we'd like to have get_size() return
the best approximation of the pickle size of the full object. So the
question is: what size is it best to have get_size() return ?

Florent

-- 
Florent Guillaume, Nuxeo (Paris, France)
+33 1 40 33 71 59  http://nuxeo.com  mailto:fg at nuxeo.com