kicad-developers team mailing list archive

[Bob Gustafson <bobgus@xxxxxxx>]

The download was bunged up, so I did a 'cd /Library/Caches/Homebrew' and 
then whacked away the kicad--bzr, the kicad--wxpython, and kicad--patch 
directories.  (This only a Cache..)

Redoing the brew install -vd kicad --HEAD step

I'm wondering if it will stumble on the patch of boost..

Waiting, waiting

Bob G


-------- Forwarded Message --------
Subject: 	Re: [Kicad-developers] Boost 1.57
Date: 	Fri, 30 Jan 2015 11:04:11 -0600
From: 	Bob Gustafson <bobgus@xxxxxxx>
To: 	kicad-developers@xxxxxxxxxxxxxxxxxxx



I'm going through with:

brew install -vd kicad --HEAD

The -vd option gives some visibility on what is happening.

If it has a problem, it halts with a request for user choice. I have 
chosen option 2 a few times so far and it is moving ahead (without 
patches though..)'

I will report how it ends up.

Bob G

On 01/30/2015 10:33 AM, Bob Gustafson wrote:
> I thought I would have a go with your tap
>
> The first time I did the brew install kicad --HEAD, it did take awhile 
> (I think it was the downloading of kicad - no progress messages 
> though). After the download, there was a file copy stage (seemed to 
> be..), but all of the file movements were conflicted with something 
> else on my disk - need to work that out.
>
> On the second time - doing brew reinstall kicad --HEAD, your patch 
> file (the boost.patch) is not finding the file it needs to patch and 
> is failing.
>
> Any guidance on what I might look for?
>
> Bob G
>
> On 01/30/2015 01:50 AM, Collin Anderson wrote:
> > On newer versions of OS X (maybe only Yosemite), building fails (and 
> > running a binary build on another, earlier system will crash on 
> > yosemite) if boost 1.54 is used, if I recall, it was something about 
> > this combo that made wxPython displeased and stop working.
> >
> > So I have been building using -DKICAD_SKIP_BOOST and linking to my 
> > own system install of boost 1.57.  This is also the strategy I use in 
> > the homebrew tap for KiCad.
> >
> > Thanks to the helpful replies from bleeding-edge enthusiasts on the 
> > http://kicad.info forums, I discovered that this was a fatal mistake, 
> > as the 'mysterious' (to me) minkowski patch was vital so boost's 
> > handling of pointers doesn't make the push and shove router crash as 
> > soon as you have to go around a corner.
> >
> > The rest of the patches have been merged with boost 1.57, so I 
> > modified the download_boost cmake module to use boost 1.57 and apply 
> > only the minkowski patch to it.  After that, I've been push and 
> > shoving my heart out without a single issue! :D That router is 
> > amazing, by the way.  I had to stop and route half a board I've been 
> > working on before I got around to making a proper branch.
> >
> > I also cleaned up the mac sections a little and removed a bunch of 
> > PPC binary checks that are no longer needed (The last version of OS X 
> > to support PPC architectures was 10.5).
> >
> > This change is only if the APPLE flag is set, it is isolated from the 
> > other platforms.  Oh, and it's agnostic to and works for an all 
> > libc++ or all libstdc++ build of wxpython and boost (10.7-10.8 use 
> > libstdc++, while 10.9+ uses libc++ I believe).  I've made a branch 
> > (~metacollin/kicad/boosting_boost) and merge request, but if anyone 
> > has issue with this, let me know and I'll do what is needed to 
> > address it.  Thanks!
> >
> > --
> > "Violence is the last refuge of the incompetent." - Isaac Asimov