Re: Standardization, large scale changes, innovations

[Manoj Srivastava <srivasta@ieee.org>]

On Wed, Mar 31 2010, Josselin Mouette wrote:

> Le mardi 30 mars 2010 à 15:42 -0700, Manoj Srivastava a écrit :
>>         I might agree that maintenance of my packages might raise a
>>  competence bar for the would-be-maintainer, and some people might fail
>>  to meet that bar.
>
> Let’s say we find this in a package:
>
> define create_md5sum
>     create_md5sums_fn () {                                         \
>         cd $$1 ;                                                   \
>        find . -type f                                              \
>               ! -regex './DEBIAN/.*'                               \
>               ! -regex './var/.*'     $(EXTRA_MD5SUM_EXCLUDE)      \
>               -printf '%P\0' | xargs -r0 md5sum > DEBIAN/md5sums ; \
>        if [ -z "DEBIAN/md5sums" ] ; then                           \
>            rm -f "DEBIAN/md5sums" ;                                \
>        fi ;                                                        \
>     } ;                                                            \
>     create_md5sums_fn
> endef
>
> Of course it doesn’t take long for a long-term developer to understand
> this is a reimplementation of dh_md5sums. However a newcomer not aware
> of your fanatic rejection of any kind of standard tools would absolutely
> not understand what this is about. And the same goes about everything
> else in the package.

        Abillity to understand fairly simple shell script is not a
 matter of tenure. It is a matter of competence.  I am dismayed that a
 fairly bland invocation of find seems opaque, in your opinion, to
 people coming into the project today. I hope that is not indeed
 true. Personally, I find a small shell snippet to be clearer than a
 reference to a external program, and when finding myself stuck in
 RHEL-land, my packages build, and would have caused more pain were they
 dependent on helper packages.  I find that justification enough.


> Now when we upgrade our policies and/or infrastructures, like what was
> recently proposed for sha1sums instead, this requires manual updating of
> all our packages for no good reason.

        That would be the naive way to implement things. And yes, that
 would be sillly.

        I just update code in one place, test it, and then run a script
 that does a git pull for all my packages. The next time I build the
 package, it will pull in the change.

> This doesn’t raise questions about the competence of the newcomer. This
> raises questions about the competence of the person who designed the
> package.

        I am happy you have an opinion. I don't think much of it, but
 you are indeed entitled to it.

        manoj
-- 
How untasteful can you get?
Manoj Srivastava <srivasta@acm.org> <http://www.golden-gryphon.com/>  
4096R/C5779A1C E37E 5EC5 2A01 DA25 AD20  05B6 CF48 9438 C577 9A1C