[PATCH modular] Perform arbitrary git or make commands.

[Gaetan Nadon]

On Fri, 2010-09-24 at 15:20 -0400, Trevor Woerner wrote:

> On Fri, Sep 24, 2010 at 3:00 PM, Gaetan Nadon <memsize at videotron.ca> wrote:
> > On Fri, 2010-09-24 at 00:15 -0400, Trevor Woerner wrote:
> > Removal of existing options should be discussed separately.
> > Providing a generic way of performing tasks (the --cmd) does not mean
> > making it harder to perform common tasks.
> 
> I'm not trying to make it harder to perform the common tasks, it just
> cleans up the code considerably.
> 

I was warning about making user interface changes based on internal code
issues.
Users look at the options to find out capabilities. They may not know
which one is a git
command and which one is a make command. We must make it easy for new
contributors.
This does not imply that what we have today is the best.


> Before I started looking into the build script, if the user specified
> the "--clean" option the script would (for every repository) perform:
> 
> 1. configure
> 2. make
> 3. make clean
> 4. make install (which essentially performs a "make" again followed by
> an install)
> 
> Also, if the user specified the "-d" and the "--clean" options the script would:
> 
> 1. configure
> 2. make
> 3. make clean
> 4. make distcheck
> 5. make install
> 
> I think it's crazy that the script chooses the order in which the
> commands get run.
> 

Now we are talking about behavior changes which needed a change in user
interface.
It's not code cleanup here, it's function change.

I had not realized the impact of the code change. I suggest you make
this function
in a separate patch.

I have some comments/information on that topic, but I will postpone.


> > Not everyone knows automake in details, much less functions like "dist" and
> > "distcheck".
> 
> If a user doesn't know enough to know what "make distcheck" is, then
> there's no way having a "-d : run make distcheck in addition to
> others" option is going to make it any clearer to them :-)
> 
> Leaving that code and those options in is almost pointless, it would
> be less work to just add the couple more make targets which are
> missing rather than implementing a generic infrastructure to run any
> arbitrary make command. If a user doesn't know what they're doing
> chances are they want "make install", which is the default. If they
> know enough to know they want to run "make distcheck" they would know
> enough to be able to supply it on the commandline. Having a "-d"
> option so the script will:
> 
> 1. configure
> 2. make
> 3. make distcheck
> 4. make install
> 
> is not efficient.


The removal of options, whether it's one or more is not related to
"Perform arbitrary git
or make commands" implies the *addition* of a new feature. It may open
opportunities
for further improvements/removal which need to be discussed.

After writing all this, I realize I was expecting a new feature while
the rest remained the same.
That explains why my comments probably did not make much sense to you.

I suggest you add the new feature, and only the new feature. Once
reviewed and in place,
it would be easier to suggest additional improvements/removal.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.x.org/archives/xorg-devel/attachments/20100924/93bfed8c/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
URL: <http://lists.x.org/archives/xorg-devel/attachments/20100924/93bfed8c/attachment.pgp>