[Mono-dev] RE: FW: [PATCH] Enum XML (de)serialization fixes

[Gert Driesen]

> -----Original Message-----
> From: Atsushi Eno [mailto:atsushi at ximian.com] 
> Sent: woensdag 8 februari 2006 9:43
> To: Gert Driesen
> Subject: Re: FW: [PATCH] Enum XML (de)serialization fixes
> 
> You keep messaging me privately, so I'd keep writing my private favor.

I'll take it to the list them, fine by me. More comments inline.

> >> -----Original Message-----
> >> From: Atsushi Eno [mailto:atsushi at ximian.com] 
> >> Sent: dinsdag 7 februari 2006 20:48
> >> To: Gert Driesen
> >> Subject: Re: FW: [PATCH] Enum XML (de)serialization fixes
> >>
> >> Hi,
> >>
> >>> Can you have a look at the attached patch ?
> >> No, I don't think the patch is ready for others' review. Too big as
> >> compared to the practical purpose, and should be split into several
> >> small patches.
> > 
> > Ok, I'll try to do that. But again, one of the reasons why 
> the patch is so
> > big is because I added numerous test cases for the changes I made.

Care to comment on this ?

> > 
> >> Lluis have applied some patches such as from Kosta,
> > 
> > I know, and my patch even includes tests for the bug fixes 
> by Kosta's patch (which did not include tests at all).

Same here ?

> >> and fixed remoting bugs those days. I don't think the reason your
> >> patch is not applied is because he is extra-busy.
> > 
> > If I'm only limited to supplying one-line fixes, then it's 
> just not possible
> > to fix any serious bugs. You must understand that.
> 
> I don't think your comment here is linked to whatever I have written.

Perhaps not, but you're referring to other patches that you think are more
likely to get committed.

> >> I'm also pretty unaware about XmlSerialization code. I even had
> >> no clue to fix some bugs reported to bugzilla, which is practical
> >> and not just for minor compatibility stuff.
> > 
> > The bugs that are fixed by my patches are definitely not just for
> > compatibility !
> 
> As long as I read the patch, the only significant but totally minor
> fix is to handle "name-escaped" items. And it could be made in
> smaller patch.

I'm sure I can split up my patch in 10 different bug reports (one of which
is #77500), each showing a significant bug.

> > Also, you explicitly asked me to look into the test 
> failures on .NET 2.0. I
> > looked into it, and fixed a few Mono test failures. 
> Ofcourse, it's not
> > always possible to fix a bug with a one-line change.
> 
> What I asked you as a System.Xml hacker was to *fix* test failures
> at least whatever you have added, which had been messing NUnit
> tests. It was rather like asking "please fix the build that you
> broke", not something else.

I know I "broke" the unit tests on .NET 2.0 (and only there). That's why I
asked for you opinion on this (through the mono-devel mailing list).

> Thus I said I'd appreciate if you would
> also take care of others as well. If you don't want, I wouldn't
> ask you to touch extra broken stuff.

Not sure I understand what you're saying here, sorry language barrier I
guess.

I definitely want to fix the other issues too. I just don't want to spend a
large amount of time on fixes that will not get committed, that's all. 

Again, I'm not trying to offend you or anyone else here: I just want to know
if I should quit trying to get fixes in svn.

> I erased a bunch of lines in the first draft to the reply to you.
> I'm largely copying it from there. I'd recommend you not to read
> more.
> 
> 
> First of all, you should ask Lluis to have a look at your patch.
> Asking others to have a look at the patch without telling the
> maintainer is highly discouraged. If I were to review the patch,
> it would have ended up to have Lluis review the patch again,
> which is totally duplicate task.

Make sense. Point taken.

> Asking anyone other than the maintainer to review the patch
> privately is highly discouraged. You could always do that in
> mono-devel-list.

I did, I just asked you if it would be possible to review the patch so I'd
know whether I can continue in the direction or not.

I asked this because I have a few days off, and I wanted to know up front if
I should bother continue working on fixing the XML serialization issues.

> I really don't understand why you keep sending private messages
> like this. It rather looks like that you always write something
> that you cannot write publicly, or you think that it is OK to
> say anything bad to me privately, which is wrong.

I'll copy my email to you here:

========================

Hi Atsushi,

Can you have a look at the attached patch ?

I've asked Lluis to review it too, but I guess he's very busy (well, you
probably too but you can't blame a guy for trying LOL).

I'd like to continue fixing issues in System.Xml.Serialization, but I've
quite a lot of changes queued locally (apart from this patch) that still
need approval, and it's getting unmanagable. I have a few days off, and I'd
like to make good use of these days (for mono).

If you have no time, no problem.

Gert

========================

Please explain: which line in this email is BAD ?

Should I start marking the "BAD" lines in your email for fun ?

> My private favor is that MS XML Serialization design is HYPE.
> It does not even handle usual structures such as choices of
> sequences (I even blogged about that for years).

Then you're saying web services are HYPE too, because they rely on XML
serialization ?

> Also, since MS XmlSerializer implementation had improved much
> in .NET 2.0, compatibility is anyways broken and it is to silly
> to mind it.

So you're saying that we should not even attempt to be compatible with .NET
2.0 ?

> Thus I dislike silly compatibility fixes, especially
> when someone asked to handle numeric values as a valid
> enumeration values (since it becomes always invalid).
> 
> One of the reason why I respect Lluis is because he takes care
> of the worst components in System.Xml. XML Schema sucks as well,
> but I maintain it because there is no other hacker that can
> handle it.
> 
> Let me summarize my private favor: I totally don't care about
> XML serialization stuff, except for test failures.
> 
> My best advice is to ask best proper person, Lluis, to review
> your patch when it got ready. He is a graceful hacker to take
> care of such mostly-for-compatibility patches.

I have no doubt on that!

Gert