kicad-developers team mailing list archive

[Chris Pavlina <pavlina.chris@xxxxxxxxx>]

On Fri, Mar 24, 2017 at 09:07:22AM -0400, Wayne Stambaugh wrote:
> On 3/24/2017 8:52 AM, Chris Pavlina wrote:
> > On Fri, Mar 24, 2017 at 08:43:00AM -0400, Wayne Stambaugh wrote:
> >> Chris,
> >>
> >> I did some testing on this and it looks good.  I have a few comments and
> >> questions.
> >>
> >> There is a noticeable response time lag when the footprints are being
> >> loaded on windows.  This may be specific to windows.  I will test it on
> >> linux when I get a chance.
> > 
> > It should only happen on first load. There's a part of the loading
> > process that I cannot make properly asynchronous. I have some ideas for
> > fixing that, which I may look into implementing, but until then the
> > first-load lag is unavoidable. The locale switching required by the
> > parser is the issue here. It's much shorter than the one when cvpcb
> > opens, though ;)
> 
> It's definitely better than waiting for CvPcb to load all of the
> footprint libraries.
> 
> > 
> >>
> >> Are you planning on filtering the footprint list?  Right now, the
> >> footprint drop down control only shows all of the footprints from the
> >> first footprint library in the table.
> > 
> > That's weird, it doesn't happen for me. Could you please give me more
> > specifics to reproduce it? I'd prefer if we could merge first - this
> > patch being so large, it's become quite a hassle to keep my branches in
> > sync while this one is up in the air. Should be a quick fix.
> 
> Type CP (assuming the device library is available) in the filter control
> and select CP1_Small.  The footprint select control contains "No default
> footprint", "Other...", and all of the footprints from the first library
> in the table (in may case Air_Coils_SML_NEOSID).  These footprints are
> all 2 pins which matches the number of symbol pins but I would expect
> all of the 2 pin footprints from every library to be in the list not
> just the first library in the table.

Could you send me the exact copy of the .lib you're using? This is what
I see: https://misc.c4757p.com/cp1_small.png

> 
> > 
> >>
> >> When cancelling the footprint viewer after double clicking on the symbol
> >> panel, the component select dialog is also dismissed.  I was expecting
> >> to fall back to the component select dialog.  I'm not sure if this is
> >> what you intended but I may not be the only person surprised by this.
> > 
> > Canceling how? If you're pressing Esc with the dropdown expanded, this
> > is one of the known UI quirks that are going to be fixed in the
> > following UI adjustments once the main patch is merged.
> 
> I'm not cancelling the component select dialog.  I'm cancelling the
> footprint viewer after double clicking the symbol panel which launched
> the viewer.  Closing the footprint viewer also closes the component
> select dialog rather than returning to the component select dialog.
> Both file->close and the close button in the title bar have the same effect.

I'm confused, how do you get a footprint viewer this way? When I double
click the symbol panel I just get the library browser.

> 
> > 
> >>
> >> It would be nice if one of our osx devs could test this before we push
> >> this to the master branch.
> > 
> > FWIW I have tested on all three platforms.
> 
> Then I'm fine with pushing this to master.  I think users will
> definitely find it useful.

Okay, I'll push it. Thanks for testing :)

> 
> Thanks,
> 
> Wayne
> 
> > 
> >>
> >> Cheers,
> >>
> >> Wayne
> >>
> >> On 3/23/2017 3:22 PM, Chris Pavlina wrote:
> >>> Hi again,
> >>>
> >>> Updated patch unless this doesn't apply cleanly anymore after b47a6e4
> >>>
> >>>
> >>> On Wed, Mar 22, 2017 at 09:09:01PM -0400, Chris Pavlina wrote:
> >>>> Hi,
> >>>>
> >>>> Footprint selection in the component chooser is now working - here is a
> >>>> patch. I'd like to merge this, but it required serious refactoring to
> >>>> make everything work cleanly, so I'm posting to the list. Wayne, please
> >>>> have a look when you get a chance.
> >>>>
> >>>> There are a couple known issues, but IMO they aren't merge-stoppers. The
> >>>> sooner I get this merged the sooner I can get actual feedback, and the
> >>>> smoother integration of any changes that are made can be.
> >>>>
> >>>> Here's a summary (yes, the *summary* is big):
> >>>>
> >>>> - DIALOG_CHOOSE_COMPONENT changes:
> >>>>    + Add FOOTPRINT_SELECT_WIDGET.
> >>>>    + Add support for DIALOG_CHOOSE_COMPONENT to pass arbitrary field value
> >>>>      overrides to the caller. This is of course to allow setting the footprint;
> >>>>      in the future it could be used to allow more field edits.
> >>>>    + Add an option to hide everything that can edit fields, for use when
> >>>>      that doesn't make sense (libedit, etc).
> >>>>
> >>>> - Add FOOTPRINT_SELECT_WIDGET
> >>>>   This is an adapter widget that combines into one FOOTPRINT_CHOICE view:
> >>>>     + Footprint listings from FOOTPRINT_LIST
> >>>>     + Filtering from FOOTPRINT_FILTER
> >>>>     + Loading progress display from FOOTPRINT_ASYNC_LOADER and wxGauge
> >>>>
> >>>>   It presents as a status progress bar that transforms in-place to a selection
> >>>>   dropdown when the footprints finish loading. The GUI remains fully interactive
> >>>>   as the footprints load; the user can even exit and reopen the dialog and it
> >>>>   will continue to load in the background.
> >>>>
> >>>> - Add FOOTPRINT_CHOICE widget
> >>>>   This is a customized wxComboCtrl with some extra features:
> >>>>     + Greying out of library name for readability
> >>>>     + List separators
> >>>>
> >>>> - Add FOOTPRINT_FILTER class
> >>>>   Provides a reusable filtered view of a FOOTPRINT_LIST, fully iterable
> >>>>
> >>>>    + Make cvpcb use FOOTPRINT_FILTER instead of providing its own filter
> >>>>      code.
> >>>>
> >>>> - Seriously rework FOOTPRINT_INFO
> >>>>    + Add partially asynchronous loading via the new FP_LIB_TABLE::PrefetchLib.
> >>>>      A FOOTPRINT_ASYNC_LOADER class is added that can spawn loader threads and
> >>>>      provide progress updates to the GUI while they work.
> >>>>    + Completely rewrite footprint loader worker threads. They are now a
> >>>>      queue-driven pool of workers rather than each loading a fixed number
> >>>>      of libs (more efficient, a bit faster) and the main thread does no
> >>>>      work, so it can return.
> >>>>    + Make FOOTPRINT_INFO available to the world, by making them virtual base
> >>>>      classes, putting the real implementation in
> >>>>      pcbnew/footprint_info_impl.h/cpp, and adding a factory function to
> >>>>      create an instance from anywhere via Kiface.
> >>>>
> >>>> - Add FP_LIB_TABLE::PrefetchLib
> >>>>   This pulls everything that is async-safe (download from github, but not
> >>>>   parsing due to a threadsafety issue) into a separate loader so the user
> >>>>   can continue interacting as footprints download.
> >>>>
> >>>>   Parsing itself remains synchronous, but the time it takes is tiny
> >>>>   compared to downloading.
> >>>>
> >>>> - Allow access to the global fp_lib_table from anywhere via kiface
> >>>>   (IfaceOrAddress()). Most methods to manipulate the table are still not
> >>>>   compiled in everywhere (they have seriously large dependencies), but the
> >>>>   table can be fetched as an opaque object.
> >>>>
> >>>> - Add a SYNC_QUEUE template class providing a std::queue wrapper with locking,
> >>>>   for ease of passing things to and from worker threads.
> >>>>
> >>>> - Minor changes:
> >>>>    + Add EDA_PATTERN_MATCH::GetPattern()
> >>>>    + Create a type to represent component history items instead of just storing
> >>>>      a list of strings and a unit. We'll need to track footprints too.
> >>>>    + Make DIALOG_CHOOSE_COMPONENT quasimodal so it can summon the footprint
> >>>>      picker.
> >>>>    + Add kiface_ids.h for storing arbitrary IDs used in kiface. This is used
> >>>>      for KifaceOrAddress(), as I'm the first person to actually use that method
> >>>>    + Remove KICAD_FOOTPRINT_SELECTOR build option, no reason for it to be
> >>>>      optional now.
> >>>>
> >>>> ===========
> >>>> Known issues, planned improvements:
> >>>>
> >>>> - History items don't remember their selected footprints.
> >>>>   There are some implementation issues and UI/UX issues I still have to work
> >>>>   out.
> >>>>
> >>>> - eeschema and cvpcb/pcbnew have separate footprint caches, so the first time
> >>>>   you open one of the latter, it'll fetch the footprints even if eeschema
> >>>>   already did.
> >>>>
> >>>> - FOOTPRINT_ASYNC_LOADER can also be used in cvpcb so cvpcb doesn't freeze
> >>>>   as the footprints load! :)
> >>>>
> >>>> - Footprint picker isn't hidden in standalone mode, but it's useless.
> >>>>
> >>>> -- 
> >>>> Chris
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> Mailing list: https://launchpad.net/~kicad-developers
> >>> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
> >>> Unsubscribe : https://launchpad.net/~kicad-developers
> >>> More help   : https://help.launchpad.net/ListHelp
> >>>
> >>
> >>
> >> _______________________________________________
> >> Mailing list: https://launchpad.net/~kicad-developers
> >> Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
> >> Unsubscribe : https://launchpad.net/~kicad-developers
> >> More help   : https://help.launchpad.net/ListHelp
>