emacs-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: New build process?


From: Tim Cross
Subject: Re: New build process?
Date: Wed, 27 Jul 2011 17:58:39 +1000

On Wed, Jul 27, 2011 at 3:08 PM, Eli Zaretskii <address@hidden> wrote:
>> Date: Wed, 27 Jul 2011 13:40:03 +1000
>> From: Tim Cross <address@hidden>
>> Cc: address@hidden, Daniel Colascione <address@hidden>,
>>       address@hidden
>>
>> As far back as I can remember, emacs
>> sources from the version control repository had an additional step
>> that had to be completed ini order to generate the  configure script.
>
> I guess your memory is either faulty or doesn't go back far enough.
> Because configure was removed from the Emacs repository only 4 months
> ago (on 2011-03-20, see the logs).
>

OK,  sorry, my error - must be confused with other projects. As has
been pointed out by others, the need to run autocont or some other
command to generate the config file is common when working from
sources directly taken from a revision control file.

>> The real issue here is whether INSTALL.BZR is an appropriate name for
>> the information that alerts people that you need to take extra steps
>> when building form sources taken from the version control repository.
>
> We had INSTALL.CVS when the VC was CVS, and I don't remember any
> complaints.
>

I thought that was what it use to be called. Partly what made me think
we had to generate the config script -  thinking again, it probably
only contained instructions relating to make bootstrap (which I don't
*think* you require for tar balls?) and some other platform specific
stuff.

Still, my main pint is I don't think we should get too carried away
trying to automate all of this. It is a common requirement and while
some may have been caught out, it is something you should expect when
working this close to the development layer. Efforts were made to
communicate the changes on this list (by you IIRC Eli) and there is
information in the INSTALL.BZR file. My objection with trying to
automate or eliminate this simple step is that the solution can often
be worse than the problem and adds just another point of potential
failure in a step which is already simple and straight-forward (once
you know about it!). However, if we can rename the file or make
another copy of the instructions under a name which the majority feel
is more likely to be noticed, great - all for that.

Tim



reply via email to

[Prev in Thread] Current Thread [Next in Thread]