Re: [Libmtp-discuss] struggling with sony nwz-e436f

[Marcus M. <ma...@je...>]

On Wed, Aug 26, 2009 at 12:35:49AM +0200, Marcus Meissner wrote:
> On Tue, Aug 25, 2009 at 06:41:45PM +0200, Linus Walleij wrote:
> > 2009/8/25 Marcus Meissner <mei...@su...>:
> > 
> > > I found another flaw in my code. The list after metadata fast
> > > reading needs to be sorted, otherwise the later code might
> > > not find stuff again (due to the binary search used).
> > 
> > OK patch applied, can you test again Nyall, if we run out of ideas
> > I'll just tag all Sony's with the bug flag to speed them up.
> > It could be that these ops are actually just slow compared
> > to getting the objinfos one object at a time, since that's how
> > Windows does it anyway.

So ... Excuse me while I put a brown paperbag on my head ...

My new code is so good in on-demand loading of information I never
noticed that the fast load method did not set "nrofobjects" in the end.

With my iRiver T10a this finally makes it work (and I actually
looked at debugging information this time ;).

(Curious enough, the T10a works for me with the "get all" method... Any
idea why it is marked not working?)

Ciao, Marcus

Index: libmtp.c
===================================================================
RCS file: /cvsroot/libmtp/libmtp/src/libmtp.c,v
retrieving revision 1.313
diff -u -r1.313 libmtp.c
--- libmtp.c	25 Aug 2009 16:38:42 -0000	1.313
+++ libmtp.c	26 Aug 2009 17:28:12 -0000
@@ -2246,7 +2246,7 @@
   }
   /* mark last entry also */
   params->objects[i].flags |= PTPOBJECT_OBJECTINFO_LOADED;
-
+  params->nrofobjects = i+1;
   /* The device might not give the list in linear ascending order */
   ptp_objects_sort (params);
   return 0;