Git Cleanup and Unfreezing
chrisj at rtems.org
Tue Aug 7 00:17:33 UTC 2012
On 7/08/12 4:42 AM, Ralf Corsepius wrote:
> On 08/06/2012 07:30 PM, Joel Sherrill wrote:
>> On 08/06/2012 12:16 PM, Ralf Corsepius wrote:
>>> On 08/06/2012 07:07 PM, Thomas Doerfler wrote:
>>>> Am 06.08.2012 19:00, schrieb Ralf Corsepius:
>>>>> On 08/06/2012 06:37 PM, Joel Sherrill wrote:
>>>>>> The general consensus has been to remove the generated
>>>>>> files from the repo.
>>>>> Whose decision? Please give the each individual's names.
> Joel, provide the names of those who decided to revert these changes.
Please add me to the list of those who want this removed.
I have tested the changes you have added and posted the results and
there are differences on MacOS to the committed source. IMO any change
is a problem.
You have not addressed the issues I raised in other threads. The major
one being the issue of a "trusted" git repo. I already do not trust it
so need to generate the files and this leads to a mess because the files
do not match and I do not know what to commit. My ability to contribute
is now limited.
I also addressed the purpose of the change in my comments. I do not see
this change helping those who actually develop RTEMS, and by that I mean
adding new code, and locating real bugs (not compiler warnings). Your
comments stated this helps those who are new to RTEMS build it. Is this
at the cost of those who are actually developing ?
I will extend this discussion to Windows testing of this change. I
cannot get any results on Windows because autoconf cannot be built .
You pushed us to this version and did not complete a full range of
testing before doing so. Being able build the tools and RTEMS does not
make them suitable for releasing. We need real test results. I suspect
your cross-compiled mingw versions are wrong. You cannot cross compile
autoconf because it is broken because it embeds the paths of the build
host in to its various scripts plus it checks the build hosts tools and
not target's tools. I worked around this years ago with the installer
exe's I made by building autoconf and automake when installing.
>>>> I have clearly voted to revert the commits in question,
>>> ... and I replied to you, trying to explain that all of your remarks do
>>> not apply.
>> So any discussion was after the fact? With no time for thoughtful
>> analysis? This decision should have had a long list of questions
>> and answers and plenty of time to answer them.
> Well, I am not commenting one details of patches addressing domains, I
> do not understand, but accept them as "the maintainer will know what he
> is doing". Now you are so bold to be wanting force me to discuss patches
> with the "clueless comments" from the mob"?
Please take a moment and collect your thought before commenting. If you
are the only person who can make changes in the build system the project
has a serious problem. Personally I do not accept we are because I for
one do know the build system and all its details so yes I can comment on
your changes. I also know all the bugs and issues that exist in it and I
choose not to comment on them because it is simpler to work around them
than deal with you.
> Pardon, but this is laughable -
No one here is laughing.
> Each person has his domains, mine are the autotools, yours
> are elsewhere.
This not an acceptable answer. All parts need peer review. All parts
need open and traceable testing.
>> BEFORE it was committed.
> See above.
See the Windows issues and the lack of testing. Sorry but the arrogance
of your comments is disturbing and your lack of testing and
documentation is obvious to all involved in this project.
>>>> since the
>>>> rational behind adding the generated files to GIT is questionable to me
>>>> and needs discussion (which did not yet occur).
>>> ... I did not reply to everybody, because I am currently on holyday and
>>> currently only infrequently read email and also do not see much sense in
>>> repeating myself in a fruitless shitsstorm initiated by people who do
>>> not want to understand.
>> IMO That reflects even worse on your judgement in committing this.
> This change was not any different from hundreds of similar patches I had
> commited over the 15 years+, I had contributed to RTEMS.
>> Even if we had discussed and agreed on this, the timing was horrible.
>> You were unavailable to address any problems accidentally caused.
> Which? I have no idea what you are talking about.
The repo is blocked while we discuss this issue. No one can commit
because of your actions. Further you commit before review and comments
can be made then head off leaving us in a frozen state.
More information about the devel