[Mono-devel-list] DataView, its very weird history and its future

[Atsushi Eno]

Hi Marc,

I don't think it is worthy of keeping my bunch of hacky fixes
in DataView (it was short term that I touched them), unless the
newer implementation is extremely buggy. Even if it was, Kosta,
Suresh and other Mainsoft/Novell guys would fix regressions.

Lately there was a bunch of code contribution from Mainsoft and
it is on the integration way i.e. could be temporarily unstable.

So let's hope that they handle things better than before ;-)

Atsushi Eno

Marc Haisenko wrote:
> Hi folks,
> while investigating DataView.cs and bugs related to it I found the history of 
> this file to be a little weird.
> 
> First a bit of background: I work for a company that works on an industry 
> project (GUI for an industrial laser) which will run on Portable .NET (aka 
> dotGNU). As you might know or not, dotGNU repackages and uses a part of 
> Mono's class library as "ml-pnet". Among other classes, the complete 
> System.Data namespace from Mono is repackaged.
> 
> We then found a bug in Mono's DataView back in fall/winter 2004, which I fixed 
> and posted a patch. A few weeks ago we hit another bug with ml-pnet 0.7.0 
> (more on that later), I checked out the current Mono SubVersion trunk, found 
> the same source as back in fall/winter 2004 and fixed the very same bug. Now 
> it turned out that the patch doesn't apply to the DataView in ml-pnet 0.7.0 
> as that is a completely different implementation.
> 
> While investigating this I found this weird history of this file (ISO dates):
> 
> 
> 
> 2004-11-11: Implementation A: I fixed a bug, posted the patch.
> 
> 2004-11-30: I reposted the patch
> 
> 2004-12-01: Miguel de Icaza commits the patch (r36917)
> 
> 2005-02-01: Implementation B: Atsushi Enomoto "mostly reimplemented" DataView 
> (r39945)
> 
> 2005-05-11: dotGNU released 0.7.0, including Mono's DataView.cs 
> (implementation B from Atsushi Enomoto, I don't know which SubVersion 
> revision)
> 
> 2005-05-16: Implementation A: Konstantin Triger overwrites DataView.cs while 
> "merging the Mainsoft branch to the trunk" with a version that pre-dates 
> 2004-12-01 (i.e. over half a year old), completely reverting Atsushis 
> implementation (r44547)
> 
> 2005-07-06: I fixed the very same bug back from 2004-11-11 and posted the 
> patch. It wasn't a applied since Konstantin was concerned that there is a 
> "deeper problem" with DataView.
> 
> 2005-07-14: Since I noticed my patch didn't go to trunk and I reposted my 
> patch
> 
> 
> 
> So my question about the future of DataView: which implementation will be 
> used ? Will implementation A be continued to be used in the future or will 
> there be a revert to implementation B (the one from Atsushi Enomoto) ? E.g., 
> do I need to bother fixing the bug in implementation B or not ;-)
> 
> C'ya,
> 	Marc
>