[Top][All Lists]

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

Re: [Axiom-developer] Maintainers

From: Gabriel Dos Reis
Subject: Re: [Axiom-developer] Maintainers
Date: 04 Nov 2006 11:00:20 +0100

root <address@hidden> writes:

| > | silver came into being in response to your requests.
| > 
| > No, I did not ask the creation of silver.  I requested you make your
| > changes public and live -- before they move to gold.  What is
| > currently called silver is not what I requested.  I did not request a
| > repository that is effectively read-only.
| > 
| > At the risk of repeating myself, let me more specific:
| > 
| >   (1) make your changes live and not just read-only.
| They're not read-only.

silver in its current state is effectively read-only:  If I commit
changes there, you don't get it automatically the next time you update
your local tree -- because you're using a different source-base. You
have to go and mangle it and have me check it what you mangled is what
I intended.


| >   (2) in particular, make it possible that we all see the patches
| >       applied (not just a message "patch --xxx" is available).
| All patches are posted to the list.

I have not seen you posting changes to the list.  I've however seen
messages saying that "patch --xxx"" is available.

| All patches to silver get automatically posted due to Bill's robot.

Only a day after, and not all changes are propagated; it seems.
One reason for sending the patches as *you* apply to silver as is that
they get more eyes to review it and test it than you would do on your
own.  Case in point is the the recent change that broke Vanuxem's code.  


| >   (3) make the master repository under SVN.  This the "main" source
| >       which which expriments, including releases are made of.
| No, SVN is not the "main" source. Arch is. SVN was created at your
| request and now you insist that it become the "main" source. This

Since I do not want to get into the revisionism business, I went to
the archive in April 2006 to see what the discussion was.

In this message:

you said:

   If any other developer feels the desire to maintain their own SVN
   branch this can also be set up on sourceforge. You can use this
   for 'public development' if you like.  

and after the creation of the SVN repository, you wrote in message:


  The SVN version is expected to be used as a hotpatch 'latest change' version. 
  The CVS version will still track the Arch. 
  On a best-effort, two-month interval I'll look to merge the changes
  from the Sourceforge/SVN to the Arch/Sourceforge CVS/Savannah CVS.

How can that be meaningfully for "public development", expected to
be used as a hotpatch 'latest change' version,  and not the main
source?  It looks to me that you just negated everything with the
recent development of silver.

I summarized the rationale of request here


   * instant availability of patches applied to mainline

   * public informative review that helps gain understanding of why
     things are the way they are, possible improvements, and things
     that should not be tried.  That is very essential to attain
     critical mass of people understanding the system.

     I've learnt a lot about Axiom through discussions between you,
     Tim and others. I think that should happen more often, on patches
     -- not just in abstract ;-)

| has not been the case for the last 5 years and won't be the case
| for the forseeable future. One key argument is that several developers
| from this list other than myself have experienced problems with SVN.

Please let be more specific about the "several developers" and discuss
what the issues really are and how they are representative and should
constitute key argument.   
>From what I've seen:

  (1) there is a problem with *SF* SVN setup
  (2) the axiom systems contain many ill-formed protability constructs
      and assumptions exposed by using SVN.

| SVN is fine for "experiments" and developers. It is NOT the release site,

For release, there should be *official tarballs*, not just --patch-xxx
under repo XYZ.


| >   (4) commit changes to the master repository before they go do gold.
| They are committed to the master repository... axiom--silver--1
| with changes that are pending in the next gold release. Multiple
| direct changes to the master tree are not appropriate.
| As an example, the most recent case of removing "duplicate" filenames

I see you have carefully constructed a very deformed idea of the patch
-- it was not about removing duplicate "filenames" as explained many

| in the debugsys image was proposed. The change seems simple and
| obvious and did not break the build-improvements branch. However, a
| careful review of the change before accepting it into silver found it
| to be incorrect.

And it was a very good thing that we had that review in public
"instantly", instead of waiting that it is published days later buried
with other changes.  The outcome of the review was a better
documentation of the relations between several components and
unwritten assumptions.  That is Good and that is why silver patches to
silver should be sent to the list and when applied made instantly
available to the community.

| >   (6) have more than just one maintainer.
| There are 22 maintainers for Arch, 6 people with complete admin access
| on savannah, and 3 people (including yourself) with complete admin
| access on sourceforge.

But, by your very words, if the main source is not SF, then it is
pointless that you have three people with complete admin access to SF
-- whether that included me or not.  That is the point.  Don't just
count.  Count right.

| Ulitmately we differ on philosophy. I believe that each branch
| has one person ultimately responsible for that branch. The responsible
| person determines policy. 

We most certainly differ on philosophy. But obviously you don't see
exactly how we differ.  So, it probably is better when you speak for
your philosophy you don't attempt to denature mine.

I believe we need a stable branch (gold), and it should be made
available, live just as any other branches.  Made live means that
changes to that branch are immediately available to everybody; instead
of being kept secret and suddenly released to the world when time
Made live does not imply that people makes applies changes willy nilly.

We should make sure that things are setup so that eventually, more
people understand the system, in particular gold and attain the point
where we don't just have one person on which everything depends.

-- Gaby

reply via email to

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