| ID: | 3796 |
| From: | ge...@igor2.repo.hu |
| Date: | Mon, 9 Mar 2020 07:10:37 +0100 (CET) |
| Subject: | Re: [pcb-rnd] new: import schematics rewrite (call for test |
| in-reply-to: | 3736 from ge...@igor2.repo.hu |
| replies: | 3797 from Majenko Technologies <ma...@majenko.co.uk> , 3798 from Gabriel Paubert <pa...@iram.es> |
Hi all, as of r29999, I've finished the following changes: - removed the old ImportGUI() action - please use ImportSch() action instead - marked Import() deprecated and made it painful; it will be removed in the next cycle; please use ImportSch() instead - removed the old Import() (gnetlist-only) menu item from File/Import; please use the more generic "Import Schematics..." menu item instead - updated all docs to reflect the removal of the old plugin Please test and report if the new import schematics feature works. Please report positive as well: if everything works - I really need to know if the new code can replace the old. The current plan is (slight change from the previous) is to keep Import() for one more cycle, with the bold warning, just for the unlikely scenario that any of our non-svn users had automated scripts using the Import() action. So I will remove the old import_sch plugin with Import() in the next cycle. I've already removed the old GUI, so users of the GUI import workflow need to switch this cycle (or with the next release). Details, from a month ago: On Sun, 2 Feb 2020, gedau@igor2.repo.hu wrote: >Hi all, > >after spending about 25 hours on this, I have more or less finished the >first version of the new import schematics feature. > >Since majority of our users work from schematics and most of them use >import schematics for forward annotation, this affects almost everybody. >You are very likely affected too. Please read this mail and the material >linked below. *** THESE CHANGES MOSTY LIKELY AFFECT YOU. *** > >This part of the code was a real dark corner. It really >needed a full rewrite. I plan to remove the old code later in this cycle >so please test the new code ASAP. > >Unfortunately the only way to clean up the mess was breaking backward >compatibility in a few places. I tried to make the transition as painless >as possible, I think it will be not that bad for most of us, but please >prepare for the following changes: > >- some menus and action names changed > >- the most common flow, gnetlist, has only a very small change and the >code usually will automatically switch over the configuration on the first >import with the new code (but we also have a GUI where you can check and >chaneg the configuration, see the links below) > >- the exotic, very rarely used make based import needs some tuning, even >in the Makefile, but we have detailed doc about that (see the links below) > >- since the old code could not import any other format or from any other >external software, the rest of the changes will affect other flows/formats >only the good way: they became available for easier, single-click import > >In return we got a much cleaner, easier to maintain code, more >modularity and major feature improvements on all parts. > > >There are three documents dealing with these changes: > > >1. Overview on what happened and why and what benefits we get: > >http://repo.hu/cgi-bin/pool.cgi?cmd=show&node=isch_overview > > >2. a howto for easy switch from the old import schematics to the new: > >http://repo.hu/cgi-bin/pool.cgi?cmd=show&node=isch_switch > > >3. a document about the new placement/removal configuration: > >http://repo.hu/projects/pcb-rnd/user/09_appendix/action_details.html#elementlist > > >The last one attempts to fix our long standing problem with how new >subcircuits placed in an import and excess subcircuits after an import >should be handled. It provides multiple options with an internal structure >that makes it easy to extend. If anyone shows active interest, I can make >script binding to this so you can write an user script for placement and >removal strategy. > >As an user I find the "frame" method of placement and "list" method of >removal the best combination. > >While being there, I've also fixed a bug we had: the nonetlist flag on a >subcircuit makes sure it is not scheduled for removal in an import. This >is a good alternative to "no refdes" subcircuits. > >Please test the new code and report any bug or srangeness. > >Best regards, > >Igor2 > > >
Reply subtree:
3796 Re: [pcb-rnd] new: import schematics rewrite (call for test from ge...@igor2.repo.hu
3797 Re: [pcb-rnd] new: import schematics rewrite (call for test results) from Majenko Technologies <ma...@majenko.co.uk>
3798 Re: [pcb-rnd] new: import schematics rewrite (call for test results) from Gabriel Paubert <pa...@iram.es>
3801 Re: [pcb-rnd] new: import schematics rewrite (call for test from ge...@igor2.repo.hu