reading that thread i am going to agree with luc-eric, create a python script/tool which creates ice attributes. the only downside is that if you are actively using that attribute (any get data node i think) the api doesn't allow you to update it. this is one current issue with crowdfx being so slow, when you need to update attributes you have to rename all the attributes you want to update to temporary names, the entire tree/graph needs to update and accept those changes. then you update the proper ice attributes and re path all the get/set data nodes to the original attribute name. if your graph is expensive you can see how this would be slow if you have to evaluate the graph twice each time this happens.
i wouldn't be worried that python would be too slow in this case because i read that file objects are just a wrapper for C stdio... http://docs.python.org/2/library/stdtypes.html#file-objects alternatively you can create a custom ice node which would do multi-phase evaluation, ie. first pass parse and cache data read from the file then re use it for subsequent evaluations. also you can use this node and 'freeze' it after the data is set to an attribute(s), just so long as you are actually using it in another graph so it doesn't get optimized out. i can't provide any actual code or direct collaboration other than advice, sorry. s On Wed, Dec 5, 2012 at 3:07 AM, Doeke Wartena <[email protected]> wrote: > hi, > > i'm really interested in a node or ppg for reading a tsv / csv file. > I often have to read this file for my study where i have to visualize data > for example. > Right now i do that in a java dialect which is quite awsome and well > suited for 2d. > 3d is also supported but i like softimage a lot more when it comes to 3d. > I had a topic about having a node for reading files on the si-community > forum: > > > http://www.si-community.com/community/viewtopic.php?f=13&t=2983&sid=78329fa63bfc4fe5076d544d5ec60c23 > > I don't have the experience and time for creating such thing. > Anyway, i was hoping someone here would like to work with me together to > create such thing. > > best Doeke >
