← Back to team overview

kicad-developers team mailing list archive

Re: Nightlies ??

 

Yes, all great reasons/comments.  Time is indeed a scarce commodity.

Fun, and feedback from your grateful users and testers are the only offset for your time.

Thanks very much.

Bob G  :-)

On 02/22/2015 05:02 PM, Adam Wolf wrote:
1) Nightly builds are going to *always* use the current revno. It doesn't make sense to do anything else. I don't even have a way to do it any differently. 2) Yes, the builds are completely automated, but supporting users isn't, and either is writing emails for this dev list. 3) There are a lot of CMake switches. Thinking of how many you have "on" vs. "off" is a non-starter. If you mean features, which features besides Python scripting aren't enabled that you're looking for? 4) Feel free to change it. If you get it tested and it works great, I might be able to switch over to it, as long as it supports easy addition and removal of patches against wx by just adding a patch to a local directory--i.e. not one in the kicad tree. You may want to wait until I outline the changes I'll be making to the scripts, or until I document it. Up to you.

Adam Wolf
Cofounder and Engineer
W&L

On Sun, Feb 22, 2015 at 4:52 PM, Bob Gustafson <bobgus@xxxxxxx <mailto:bobgus@xxxxxxx>> wrote:


    On 02/22/2015 03:44 PM, Adam Wolf wrote:
    Any patches in my nightlies against kicad are for packaging
    purposes only.  I do not believe there are any now.  This is why,
    for example, I did not include the trackpad support in the
    nightlies, even though I run it on all my personal builds.

    (Will it be easy for me to *also* provide a trackpad branch? Yes,
    but nightlies are only a few days old, and I need a little break :) )

    Yes, thanks much for your efforts to bring the OSX platform up to
    the level of Linux and Windows. It is not an easy job and you have
    been miraculously diligent.


    With regards to boost--I have no idea why people want to run on a
    newer point release of boost on KiCad, especially on Mac.  I
    remember the bug reports, years ago, of super weird behavior due
    to issues in boost, and the complicated efforts it took to
    isolate the issue to boost, and to fix them.  It isn't like we're
    using an ancient version of boost or anything, either.

    I think the need for 1.57.0 was to support the python scripting. I
    may be wrong about this - haven't clicked through the emails on that.


    The fact that we're still getting weird bug reports (like the one
    today or yesterday) makes me want to be *more* conservative with
    my nightlies, not less.
    Perhaps a conservative approach would not show the problems. I'm
    assuming that by conservative you mean using a lower revno, or
    less switches 'on'.

    New test builds should show new problems (assuming that the old
    problems have been fixed).

    A failing build image is a motivation!

    Every minute I spend on support is a minute I'm not writing new
    features (like a default fp-table-lib selector so users who want
    to use a "normal" configuration will be able to use them without
    copying files around), helping get features other people wrote
    ready to merge into master (like the trackpad support), enhancing
    the nightlies, (adding python scripting support, adding a
    build.log in the dmg, or providing a changelog that shows a list
    of commit messages between each night's versions), or working on
    automated testing to help maintain or even increase the quality
    of KiCad.

    Yes, of course - but you now have the nightlies automated, yes?
    Automation means no thought required... :-)

    There's no magic about the builds, by the way--every night,
    around midnight CST, they basically check out the latest source,
    and run through the mac-osx compiling document.  I use my own wx
    compiler script, but it basically does the same as the one in the
    tree--I just wrote mine before there was one in the tree.

    The scripts are available here:
    https://code.launchpad.net/~adamwolf/+junk/kicad-mac-packaging
    <https://code.launchpad.net/%7Eadamwolf/+junk/kicad-mac-packaging>  I'm
    not really happy with them, and I want to rewrite most of them,
    but doing that is relatively low on my list.  I also haven't
    really publicized these, because I'm worried people will use them
    as general purpose OS X builders, and there's some optimizations
    and tradeoffs made in order to make it easy to integrate into the
    Wayne and Layne build cluster.  It works on a regular system,
    mostly, but any existing documentation on it currently assumes
    you know what you're doing, mostly. :)
    I would prefer that the tree script be used, but I haven't
    compared the two to see if it matters at all.


    Adam Wolf
    Cofounder and Engineer
    Wayne and Layne, LLC

    On Sun, Feb 22, 2015 at 1:38 PM, Bernhard Stegmaier
    <stegmaier@xxxxxxxxxxxxx <mailto:stegmaier@xxxxxxxxxxxxx>> wrote:

        Well, I think the point is to have KiCad at bleeding edge,
        but not every dependency library.
        As long as there are no known issues with an old boost, et.
        al., why should you want to be also at bleeding edge of those?

        Updating everything frequently just makes it harder to see if
        issues are in KiCad or in an updated dependency.

        And wrt to features Adam already told that he will be working
        on providing a scripting enabled build next.


        Regards,
        Bernhard

        > On 22 Feb 2015, at 20:31, Bob Gustafson <bobgus@xxxxxxx
        <mailto:bobgus@xxxxxxx>> wrote:
        >
        > I am thinking that the point of 'Nightlies' is to build at
        the 'bleeding edge' and provide the builds to folks who can
        then easily try their normal working 'use case' to see if
        things (still) work. If they crash - fine, we have a data point.
        >
        > Looking at the Version info of the current 'Nightlies', I
        see boost still at 1.54, and many of the environmental
        variables turned 'off'. The source code version does seem
        pretty close to the repository revno, but do the added
        patches affect things that are not turned 'on' by the
        environment variables?
        >
        > Just a thought.
        >
        > Bob G
        >
        > _______________________________________________
        > Mailing list: https://launchpad.net/~kicad-developers
        <https://launchpad.net/%7Ekicad-developers>
        > Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
        <mailto:kicad-developers@xxxxxxxxxxxxxxxxxxx>
        > Unsubscribe : https://launchpad.net/~kicad-developers
        <https://launchpad.net/%7Ekicad-developers>
        > More help   : https://help.launchpad.net/ListHelp


        _______________________________________________
        Mailing list: https://launchpad.net/~kicad-developers
        <https://launchpad.net/%7Ekicad-developers>
        Post to     : kicad-developers@xxxxxxxxxxxxxxxxxxx
        <mailto:kicad-developers@xxxxxxxxxxxxxxxxxxx>
        Unsubscribe : https://launchpad.net/~kicad-developers
        <https://launchpad.net/%7Ekicad-developers>
        More help   : https://help.launchpad.net/ListHelp






Follow ups

References