Hi Juergen, I haven't seen anything about xcu files. I just zip up the META-INF files, the python script, and the binary version of my IDL file.
Do you have a pointer handy to the newest format? thanks joachim On 5/25/07, Juergen Schmidt <[EMAIL PROTECTED]> wrote:
Hi Joachim, glad to hear that your add-in works now. Well not everything was possible but you are able to workaround the multiple returns. I would suggest that you take a look on new add-in with a configuration file to define the new add-in function 8with localization if you want). This new approach has several advantages and it would be nice if you can support this. I know it is not so important for you because your add-in works but i think it is worth to support the latest features at least for new implementations. Or did you already have an xcu file for your add-in? Juergen Joachim Dahl wrote: > Hi Andreas, > > I think the Python UNO bridge works fine; I managed to return an array of > values by using an array function. > I needed to return multiple arrays of different lengths; to give a > specific > example I made a wrapper around a > linear programming solver with input arguments 'c' (a vector), 'G' (a > matrix) and 'h' (a vector). This routine > outputs a status string, a vector 'x' and a vector 'z', and I used an > IDL > as > > string lp ([in] sequence < sequence <double> > c, > [in] com::sun::star::table::XCellRange x, > [in] sequence < sequence <double> > G, > [in] sequence < sequence <double> > h, > [in] com::sun::star::table::XCellRange z); > > i.e., I used the XCellRange as a way to return additional variables by > "writing back" values in x and z. This approach > works well for me in Python. > > The only problem I have left is that I have optional arguments, some of > which are also "output" argument (albeit, in > the sense of being XCellRanges), but that's not possible with a Calc > add-in. So my work-around for this problem > is to have a different function 'lp2' with more arguments (all mandatory). > > But in summary, the Python UNO bridge seems fine for writing add-ins. > > Joachim > > > > On 5/25/07, Andreas Saeger <[EMAIL PROTECTED]> wrote: >> >> Joachim Dahl wrote: >> > me. >> > >> > So my problem is not so much of how to write a prototype of an >> addin, but >> > rather how to >> > implement something more advanced like "methodFour". >> > >> > Joachim >> > >> >> Hi Joachim, >> http://wiki.services.openoffice.org/wiki/CompleteAddIn has a chapter >> "Transforming sequence of sequence into array". This describes the same >> problem as I tried to explain with my simple Basic function TWICE(array) >> in http://www.oooforum.org/forum/viewtopic.phtml?t=57196. Only >> array-functions, such as the built-in LINEST, can return more than one >> value. The resulting array has to be converted to a 2-dimensional matrix >> rather than a list of lists. So it is possible that python alone is not >> suitable for add-ins that deal with sheet functions in array context. >> >> Andreas >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [EMAIL PROTECTED] >> For additional commands, e-mail: [EMAIL PROTECTED] >> >> > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]