On Thu, 2007-03-15 at 06:37 -0500, Dan McMahill wrote: [snip]
> The next step is for interested students to apply to Google. You may > either select a project from our suggestion list, or come up with your > own ideas for a project. [snip] Peter B. and myself have taken a look over the proposed projects, and have some comments: Project Manager --------------- Some mention of xgsch2pcb may be useful as a starting point. There is no gschem<->PCB DBUS interface... Peter C added a DBUS interface to the "actions" mechanism in PCB, but there is a long way before this is a full IPC solution. Lock files may not be the best solution for concurrent schematic access. Kernel level locking should be investigated - as well as library level shared state. Look at man 2 fcntl (Advisory locking etc..) Spice improvements ------------------ See notes on DBUS support (above). We need to define an API to expose for gschem. (Related to the forward / back annotation project, and other libgeda changes). Parts manager ------------- We spent some time discussing details on this, but don't have any solid suggestions yet. Gnetlist/gnetman support for hierarchy -------------------------------------- This doesn't seem suitable for a SoC project. We may be wrong, but isn't this at the very core of GAF + libgeda? We are both keen to see native netlist database support in libgeda to facilitate integration with other tools. (Retaining gnetlist and its scheme backends). This should probably all be integrated with any libgeda improvements being made.. i.e the next project: libgeda API formalisation ------------------------- Presumably based on suggestions by Peter B, however we feel that the difficulty level is much understated - as it will require intimate familiarity with _all_ of the GAF code-base to start. Mention was made of a guile API.. presumably underlying this is a C-API, which will require a decent amount of work in its self. Perhaps we could leave the language bindings for a later project? Make gsch2pcb use (same search paths as) PCB -------------------------------------------- The second part of this project is perhaps the way forward.. using a defined API to make PCB write its own files. This meshes nicely with interactive forwarding of changes to the layout (e.g. xgsch2pcb or a project manager.) gwave ----- Both Peter B and myself have used GTKWave, and wonder if this could be extended to support new and existing analog only data-formats. (It already support viewing an analog representation of a binary vector). It can view as discrete steps, or interpolate. It also appears well equipped to handle large data files. PCB Auto-router --------------- The student might like to look at making the auto-router pluggable (e.g. HID), and then investigate porting the "MUCS" auto-router as a replacement option for the current router. http://www.cs.manchester.ac.uk/apt/projects/tools/mucs-pcb/ Improved forward / back annotation ---------------------------------- This is likely dependant on the API improvements / changes mentioned else-where - and thus might be a difficult project to run concurrently with some of the others. Peter Clifton Peter Brett _______________________________________________ geda-dev mailing list [email protected] http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev
