X-Authentication-Warning: delorie.com: mail set sender to geda-user-bounces using -f X-Recipient: geda-user AT delorie DOT com X-Injected-Via-Gmane: http://gmane.org/ To: geda-user AT delorie DOT com From: Kai-Martin Knaak Subject: Re: [geda-user] work on gEDA Date: Sun, 01 Mar 2015 21:41:25 +0100 Lines: 99 Message-ID: References: <20150215021721 DOT 28d94fde AT jive> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7Bit X-Complaints-To: usenet AT ger DOT gmane DOT org X-Gmane-NNTP-Posting-Host: a89-182-175-16.net-htp.de User-Agent: KNode/4.14.1 Reply-To: geda-user AT delorie DOT com Lev wrote: > Please everybody write down a wish list of PCB and gschem. We shell > select what is the most important issue and start making it right. Oh, I got lots of wishes! Needed some time to formulate and prioritize. I confined myself to a list of ten (plus a zeroth one): * My number one wish would be to regain the ability to do a windows version of the geda tools. It used to be possible to cross compile the whole suite. A PhD student at my day job even prepared an installer, so the geda suite could be installed like any other major piece of software. Unfortunately, his tool chain broke when geda started to require guile >2.0 . Since guile 2.0 and onwards the cross compile suites fail to produce a working windows version of guile. IMHO, this is a major regression. It puts geda behind its main contenders kicad and eagle. Irritatingly, the guile developers do not seem to seems to care. I got zero response on their mailing list :-| * My number two wish would be scripting, both in pcb and in gschem. That is, a way to do everything the GUI does but in a script. geda-gaf and pcb both have implemented some aspects but not enough to actually make it useful in the sense my wish calls for. See how freeCAD uses python or the way eagle uses "ulp" scripts. In pcb the concept of "actions" goes a long way. But there is no glue to use them conditionally or iterate in loops. Everything that involves explicit mouse input is missing from the set of actions. In geda-gaf the situation is somewhat complimentary. There is algorithmic glue in the form of the ability to interpret guile scripts. But unless I missed something actions are missing and there is no way to make the script put a symbol somewhere. * My number three wish would be blind and buried vias for pcb. Blind and buried vias are vital when it comes to highly integrated digital components. Most significantly, it blocks me from using FPGAs except for the smallest versions. Blind and buried vias is a larger request than it looks. It requires a change of pcb format. When done right, it involves a redesign of the way layers are represented internally. * My number four wish would be EDIF import/export both for pcb and for gschem. This would open doors to all kinds of work-flows, including interaction with other EDA suites. * My number six wish would be better integration of simulation in the GUI. Running a simulation of a simple circuit should be as straight forward as making a PCB of the circuit. This requires proper handles in the GUI and a decent set of models in the default library. See qucs or LTspice for examples how set-up of a simulation can be done in a user friendly way. * My number seven wish would be the notion of packages. A package would be a container for all information that are associated with a component: Symbols, footprints, values, variants, SPICE models, 3D models, where to buy, notes... Note, the plural. A component might be represented by more than one symbol. There may be different footprints (DIP8, SO8, SOT23-5, ..), or different values (10nF, 12nF, 15nF, ..). Symbols, footprints and models can be directly contained in the package or they can be referenced by a their name. Both approaches have their pros and cons. It is up to the maintainer of a library to draw the line and decide what exactly is included in a package of her library. Packages would be referenced in schematics much like symbols are now. Unlike now, the package "knows" about alternative footprints and values. So the GUI can present list to the user to choose from. Packages would also be referenced in the layout. This helps to keep layout and schematic consistent. It would facilitate back-annotation. Packages would be the atoms of a geda library. The notion of packages is quite common in EDAs (see eagle, or protel). IMHO, quite a bit of the awkwardness with geda libraries originates from its absence. Packages would allow for a scalable and easily sharable libraries. * My number eight wish would be the concept of groups both in gschem and in pcb. * My number nine wish would be shove and push of lines in pcb. I used this intensively in protel99. It made manual routing of dense layouts so much less tedious. * My number ten wish would be a GUI that gravitates a little more toward the concepts that have emerged and proven useful in the rest of GUI driven applications. From the top of my head: tabs to deal with several open documents. configurable toolboxes in-place-editing of strings visible handles at sensitive places of a focused object a work-area for user input rather than a multitude of pop-ups some kind of feedback on the object a mouse click would refer to And then there is zeroth wish for better documentation. The current state can be characterized as a "historically grown pile" in various formats and various states of up-to-dateness -- This is neither attractive to the new users nor always useful to the experienced. ---<)kaimartin(>---