[csw-maintainers] Buildfarm setup - documentation

Peter FELECAN pfelecan at opencsw.org
Sat Oct 5 13:48:42 CEST 2013

Maciej Bliziński <maciej at opencsw.org> writes:

> On Sat, Oct 05, 2013 at 12:24:40PM +0200, Peter FELECAN wrote:
>> Maciej Bliziński <maciej at opencsw.org> writes:
>> (...)
>> >> sqlobject.dberrors.OperationalError: Row size too large. The
>> > maximum row size for the used table type, not counting BLOBs, is
>> > 65535. You have to change some columns to TEXT or BLOBs
>> >  
>> > Hm. This happens when you try to initialize the database and create the
>> > tables, not when you try to insert any data.
>> >
>> > Is this the old code or the new code?
>> >
>> > If it's the new code, then it's a bug. But I haven't been able to
>> > reproduce it.
>> >
>> > If it's the old code, then it's what I mentioned earlier: It's not the
>> > instructions we don't have, it's the code we don't have.
>> revision 22118. consequently what I imagine you call "new code".
> No, the new code is still in git, on github. If it's from subversion,
> it's the old code.

Good. I think that at this point we must clarify things:

1. I volunteer to verify the manual in what concerns the build-farm setup.
2. You write and consolidate the pieces for a draft
3. I'm testing and adapting the manual and we arrive at a point where
   the procedure doesn't work.
4. You explicitly write about .buildsys/v2 in mgar in the draft as
   in our exchanges
5. Now you are writing about a "new" code which is in git.
This is quite confusing. I wonder what is the reason.

Also, when asking you if it's a good idea to package the required
tool you say:

1. That using "svn up" is not a really difficult
2. But, later, you say that it's in a git repository
3. You give a list long as a fasting day for why packaging the
   checkpg related tool is not a good idea

This is also very confusing.

Can you, please, explain your standpoint?

Thank you in advance

More information about the maintainers mailing list