launchpad-dev team mailing list archive

[Gary Poster <gary.poster@xxxxxxxxxxxxx>]

On Aug 3, 2009, at 12:08 PM, Maris Fogels wrote:

> Hi everyone,
>
> So I've noticed some people have had problems with the initial  
> build, and the first thing they try is 'make clean'.  However, this  
> doesn't do what they expect: instead of resetting the build, it just  
> resets the application state. Subsequent attempts to use 'make'  
> still fail, because the build artifacts are still broken.
>
> So, to make things easier for our developers, should we follow a  
> more standard set of behaviours for our Makefile targets?
>
> From the GNU Coding Standards: Standard Targets:
> http://www.gnu.org/prep/standards/html_node/Standard-Targets.html#Standard-Targets
>
> 'clean'
> Delete all files in the current directory that are normally created  
> by building the program. Also delete files in other directories if  
> they are created by this makefile. However, don't delete the files  
> that record the configuration. Also preserve files that could be  
> made by building, but normally aren't because the distribution comes  
> with them.
>
> 'distclean'
> Delete all files in the current directory (or created by this  
> makefile) that are created by configuring or building the program.  
> If you have unpacked the source and built the program without  
> creating any other files, ‘make distclean’ should leave only the  
> files that were in the distribution.
>
> The standard defines 'mostlyclean' and 'maintainer-clean' targets as  
> well.
>
> As an example of how this would help, when someone tells us "It says  
> I have an error with foo.egg", we can simply tell them to run 'make  
> distclean; make', and everything will be reset as expected.
>
> Should we make this change?

I'm not thrilled with adding distclean, though I experienced one  
example of the use case this morning with Jono.  Someone had  
inadvertently broken an egg (see bug 408413, https://bugs.edge.launchpad.net/launchpad-foundations/+bug/408413) 
.  I have not seen any other problems like this.  Any other stories  
out there?

We share our eggs across builds, just as we share our sourcecode.   
Therefore, running make distclean would break any branches using  
different eggs until they too reran make.  That's surprising.

I like sharing eggs; it saves a lot of build time.

I suppose I'd be ok with a distclean target that simply added a  
warning to  the output that it may make it necessary to rerun make in  
other branches.

Rereading the description of ``distclean`` in the quote above, perhaps  
a somewhat more elegant solution from a user perspective would be for  
a branch to remember what eggs it created, and just clean those.   
That's still fragile for considerably more work, so it doesn't strike  
me as a win.

So, in sum, maybe, but beware. :-)

Gary