Recent changes to feature-requests

slice feature wanted for Array3dC

2003-09-10T16:37:08Z

It would be nice if there was a method for Array3dC
that returned an Array2dC handle to the ith slice, as
discussed

Cheers,
Bill.

image IO feature request

2003-08-28T16:48:01Z

Would be nice to have:
o pbm format handling
o ImageC<bool> handling

Cheers,
Bill.

Macro for definition of typenames for IO system

2003-07-22T11:19:46Z

As I understand for each class which should support serialization it is
necessary to define set of static variables. For example

KernelFunctionC LinearKernel2KernelFunction(const LinearKernelC
&func)
{ return func; }

DP_REGISTER_CONVERSION_NAMED(LinearKernel2KernelFunctio
n, 1, "RavlN::KernelFunctionC RavlN::Convert(const RavlN::
LinearKernel2KernelFunction &)");

static TypeNameC TypeLinearKernel(typeid(LinearKernelC), "RavlN:
:LinearKernelC");

FileFormatStreamC<LinearKernelC>
FileFormatStream_LinearKernel;
FileFormatBinStreamC<LinearKernelC>
FileFormatBinStream_LinearKernel;

It seems that it would be good to have macros which will help to do
that. Moreover there are lots of reasons to define it apart from class
defining simplicity.

virtual files for conv

2003-07-21T17:01:40Z

I would like to use conv to grab stuff from a camera &
dump it to a file (not real-time). But conv does not
understand '@V4LH'.

Cheers,
Bill.

Audio extraction from dv frames.

2003-07-21T13:14:06Z

Please could audio be added to conv so that I can get
the audio out of a dv file, e.g.

conv file.dv file.wav

Cheers,
Bill.

large file handling

2003-07-11T08:28:50Z

Please could VPlay - and preferably any video i/o -
handle files >2GB without fuss. doDvGrab already does
this, but Lee had to put special stuff in the main prog.

Cheers,
Bill.

XML DTD handling would be nice

2003-06-30T13:31:05Z

Under XML Streams it says:

......... If there is demand for this functionality
[i.e. DTD handling] a RAVL wrapper class for one of
these libraries may well be added to the RAVL Contrib
code base.

May I express such a demand? Somelthing along the
lines of what Kieron did for ASSAVID would be really nice.

Cheers,
Bill.

HashC / RCHashC documentation

2003-06-17T14:00:37Z

Can't find my last whinge on this one - it seems to
have vanished (do I hear you all saying "just as well"?).

Please could the general description of these 2 classes
be improved. In particular:

1) Why does HashC exist at all as a
non-reference-counted object? - it seems to be contrary
to RAVL tradition. When would I want to use a
non-reference-counted hash table?

2) Maybe the documentation of the 2 classes could be
made more similar (except of course where the classes
themselves differ)

3) Make it clear that the extra functions needed refer
to the key, not the hash table

4) Tidy up comments beginning with "!!!!": it is not
clear whether something is missing from HashC, or
wheter this is yet another thing required of the key.

Hope this doesn't all disappear down the plughole this
time.

Cheers,
Bill.

@X virtual file problem(s)

2003-06-09T11:05:29Z

Main problem:

My application opens 5 @Xs. If I am slow with the
mouse to locate them on the screen (this happens as you
get older), the whole program hangs permanently. If I
am snappy about it, it is ok.

Small problem:

The info field seems to update when I move the mouse
(fine). It also needs to update when the display updates.

Query:

How can I temporarily halt the application (e.g.
ctrl-Z) without freezing the @X display?

Cheers,
Bill.

New features for iterators

2003-05-14T17:02:43Z

I have 2 problems with the iterators (well, the 2D ones
anyway):

1) When the constructor bombs out because the
rectangles don't match, there is no helpful message
telling you why - you just get:
Ravl assertion failed
/vol/vssp/localsoft/Auto/linux/trees/Ravl.1/src/Ravl/Core/Container/Array/Array2dIter3.hh:49

Please could they have a helpful message saying that
there was a rectangle mismatch.

2) It would be much more useful if the assertion failed
only when the first rectangle did not fit inside the
others. Currently it requires an exact match, which is
not necessary.

Cheers,
Bill.