This is the mail archive of the
mailing list for the GCC project.
Re: PATCH for Re: contrib/gcc_update
- To: Gerald Pfeifer <pfeifer at dbai dot tuwien dot ac dot at>
- Subject: Re: PATCH for Re: contrib/gcc_update
- From: Richard Earnshaw <rearnsha at arm dot com>
- Date: Wed, 23 May 2001 16:24:08 +0100
- cc: Phil Edwards <pedwards at disaster dot jaj dot com>, Mark Mitchell <mark at codesourcery dot com>, Brad Lucier <lucier at math dot purdue dot edu>, Tom Tromey <tromey at redhat dot com>, gcc at gcc dot gnu dot org, Bryce McKinlay <bryce at waitaki dot otago dot ac dot nz>, Richard dot Earnshaw at arm dot com
- Organization: ARM Ltd.
- Reply-To: Richard dot Earnshaw at arm dot com
> On Mon, 7 May 2001, Gerald Pfeifer wrote:
> > I think we should make -d a default option in gcc_update and will make
> > that change in a few days.
> > Those that don't really need/want the entire gcc CVS module can easily
> > change the script (and probably don't use it anyway), while this change
> > will make the default case work more as expected by regular "users".
> Especially given that this has bitten a few more of us recently (Mark,
> David,...) I have now installed the following patch on mainline:
> 2001-05-23 Gerald Pfeifer <firstname.lastname@example.org>
> * gcc_update (UPDATE_OPTIONS): Add -d to the default settings.
> Mark, I have not installed this on the branch, under the assumption that
> we won't add new directories there and thus avoid this issue altogether.
> Please let me know if you prefer otherwise.
I some issues with this
1) It's impossible to turn it off
2) It increases the traffic during the update (sometimes I *know* there
are no new directories) -- When I'm updating over a dialup this can
increase the time for the update considerably, particularly if I have
conflicts that need sorting out and are then followed by a second update
3) Using -d on a branch without -r branch_name can cause the new
directories to get pulled from the trunk rather than the branch (at least
this is what used to happen on some older versions of CVS, it might have
been fixed by now).
I think having a solution to 1) should make the other two insignificant,
but I think without such a solution this is not a good idea.