RE: [PATCH] Revised: OneFS build issues for Samba 3.2 - Samba

This is a discussion on RE: [PATCH] Revised: OneFS build issues for Samba 3.2 - Samba ; Sorry, Tim noticed an issue in my patches: I was still based on v3-2-test. I just moved over to v3.3 and found s'more build issues on our platform, will try to have a better patch set out later tonight. > ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: RE: [PATCH] Revised: OneFS build issues for Samba 3.2

  1. RE: [PATCH] Revised: OneFS build issues for Samba 3.2

    Sorry, Tim noticed an issue in my patches: I was still based on
    v3-2-test. I just moved over to v3.3 and found s'more build issues on
    our platform, will try to have a better patch set out later tonight.

    > -----Original Message-----
    > From:

    samba-technical-bounces+zachary.loafman=isilon.com@lists.samba.org
    > [mailto:samba-technical-
    > bounces+zachary.loafman=isilon.com@lists.samba.org] On Behalf Of

    Zachary
    > Loafman
    > Sent: Thursday, July 03, 2008 11:47 PM
    > To: samba-technical@lists.samba.org
    > Subject: [PATCH] Revised: OneFS build issues for Samba 3.2
    >
    > Ignore the ones below. Better patches attached.
    >
    > The first two are configure changes:
    > 0001 - Check for f_frsize when you find statvfs.
    > 0002 - Adds an --enable-picky-developer to add -Werror for those of

    that
    > really, really don't like warnings. (BTW, thank you so very much for
    > getting 3.2 to compile as clean as it does, it was a real pain getting
    > 3.0 to compile with -Werror.) If you don't like this change, I'll
    > happily just make sure it's a CFLAG for us, but I'd like to encourage
    > people to be pickier.
    >
    > The next two couple be merged or stays separate:
    > 0003 - Fixes various general warnings. The only potentially
    > objectionable one here is the iconv() change. If you wish, I can add a
    > configure test for whether the second argument of iconv() is properly
    > constified (on BSD it is, but presumably whoever added the (char **)
    > cast has a platform where it is not), but given that we wrap it
    > immediately and then don't use the native iconv elsewhere, it seems

    fine
    > to (void *) it.
    > 0004 - Fixes some warnings that I'm pretty sure are specific to

    FreeBSD.
    > Our 6.1 based system has reboot defined in unistd, so any variable

    using
    > the name tickles -Wshadow, similarly for tcp_seq.
    >
    > After these, I can build without warnings on our platform, yay.
    >
    > ...Zach
    >
    > > -----Original Message-----
    > > From:

    > samba-technical-bounces+zachary.loafman=isilon.com@lists.samba.org
    > > [mailto:samba-technical-
    > > bounces+zachary.loafman=isilon.com@lists.samba.org] On Behalf Of

    > Zachary
    > > Loafman
    > > Sent: Thursday, July 03, 2008 12:29 PM
    > > To: samba-technical@lists.samba.org
    > > Subject: [PATCH] OneFS build issues for Samba 3.2
    > >
    > > Attached are patches for two build issues that occurred when

    > attempting
    > > to build Samba 3.2 on one of our nodes. We're roughly FreeBSD 6.1

    > based,
    > > so I wouldn't be surprised if they appeared there as well.
    > >
    > >



  2. Re: [PATCH] [Re-Revised]: OneFS build issues for Samba 3.2

    Hi, Zach!

    On Wed, Jul 9, 2008 at 9:19 AM, Zachary Loafman
    wrote:
    >
    > In addition, I rebased all the previous patches:
    >
    > 0001 - Check for f_frsize when you find statvfs. Yes, this is an oddity
    > of our platform.


    Actually, are you sure, that stavfs() should be used? If you platform
    based on FreeBSD - I'd rather use stafs().

    > 0003 - Fixes various general warnings. The only potentially be
    > objectionable one here is the iconv() change. If you wish, I can add
    > configure test for whether the second argument of iconv() is properly
    > constified (on BSD it is, but presumably whoever added the (char **)
    > cast has a platform where it is not), but given that we wrap it
    > immediately and then don't use the native iconv elsewhere, it seems fine
    > to (void *) it.


    This seems duplicate the work, been done in the FreeBSD samba3 port,
    dunno, why this still wasn't comited into the main tree

    > 0004 - Fixes some warnings that I'm pretty sure are specific to FreeBSD.
    > Our 6.1 based system has reboot defined in unistd, so any variable using
    > the name tickles -Wshadow, similarly for tcp_seq.


    This ones would be nice to get commited too...

    I'll try to check your patches together with other FreeBSD patches to
    see, how do they fit the general distribution.

    Thank you for sharing the code!

    With regards,
    Timur.


+ Reply to Thread