Re: Please hold off modifying sc in the next few weeks

On 06/14/2013 10:11 AM, Bjoern Michaelsen wrote:
> Hi Kohei,
>
> On Fri, Jun 14, 2013 at 09:29:56AM -0400, Kohei Yoshida wrote:
> > I'm in the middle of a huge (I mean huge) refactoring in Calc core,
> > which will take at least in the next few weeks to complete.  It
> > touches pretty much every piece of the code in ScColumn and many
> > surrounding code.  So, I'd appreciate it if you guys could hold off
> > changes such as German comment translations and string conversions
> > etc until this work is done. Having these other changes would only
> > make my work a lot harder and I would have to spend more time
> > merging, which delays completion of this work.
> Im not too happy with essentially shutting down all development on master on
> one application for an undefined timeframe. Can we make this a bit more
> bearable by:
I'm not shutting down all development. If you know what you are doing 
and know that your change won't affect my ScColumn cell storage rework, 
feel free to push. Also, if your change is urgent, and/or you can't 
sleep at night until you push your change, go ahead and push.  That 
means I instead may have to stay up late a little, but I guess nobody cares.

> a/ a concrete timeframe, I would suggest until the 2013-06-30
>     (with the option to extend the block another week a few days before the end
>     should the need arise)
I can't give any more concrete time frame than the next few weeks, since 
even *I* don't know exactly when it will finish.
> b/ make that work trackable and visible on a regularly pushed branch (maybe
>     even regularly rebased), so it easier to see where possible conflicts arise
I don't want to do that because this is a huge change that won't even 
compile. I've been working on this in the past 3 to 4 weeks without 
being able to build it.  And as I understand it some people (especially 
those who admin bibisect, tinderboxes etc) do want to make sure that 
each commit is at least buildable, if I'm not mistaken.

But if you really want to see what's going on, track my change on

feature/formula-core-rework

at

ssh://anarchy.freedesktop.org/~kohei/libreoffice/core

Right now I can't afford to make each commit meaningful, and each commit 
is not in a buildable state. What I'll do is to squash all these 
fragmented commits into one once I finish all necessary changes, make it 
buildable, and make it pass all existing unit tests, before merging to 
the main repo.

> c/ instead of suggesting to stop all work on calc, divert those changes to
>     gerrit, were those could stay "on hold" until the refactoring is done, and
>     be merged in due time afterwards ...
Fine. Then, my request is for gerrit reviewers.

> Note also that some string refactorings or friends are global (e.g. removing a
> member function), and excluding these would block all of LO, which is obviously
> not a good thing.

Whether it's a "good thing" or not, I'm simply requesting some 
assistance to make my life a bit easier (as it has been rather brutal in 
the past few weeks). If you think your change is important, feel free to 
push then.  I'll have to suffer a bit more, but I guess I'm not a 
priority in the whole scope of this project, and I'm fine with it.

Or, if none of these make sense, or you find this confusing or too 
restricting, then just ignore my request and go on with your lives.

Peace,

Kohei

--
Kohei Yoshida, LibreOffice Calc hacker, SUSE.