On Tue, Jul 11, 2006 at 10:54:01PM +0200, Johannes Winkelmann wrote:

...

...

If you think that w/o splitting and debugging changes the patch would be far smaller, you are very wrong. I don't really have an interest in discussing this to death. We just expect large patches to be split into smaller patches, each introducing a single change (pretty much like the linux kernel development works).

If you submit a reasonably split set of patches we'll definitely take the time to review and discuss it, or point out when a patch violates the "one patch, one feature" rule.

[typedefs]

...

...

Well, if I see std::pair<string, string> file;

You got error here: it's pair<string,fileinfo_t>. With typedef you are safe from errors like that.

In fact I merely wrote that as an example, however you instantly spotted a potential error since an explicit type was used instead of a annonymous typedef. Or to paraphrase you, "with typedef such errors are hidden from you".

So bottom line is there's no "better" here, just preference, and you may want to take the maintainers' preference into consideration.

Wait. Maintainer prefer code like: compose1( first_in_pair<pair<string, fileinfo_t>, string>(), first_in_pair<pair<string, fileinfo_t>, conflict_t>, pair<string, fileinfo_t>() );

and

void pkgutil::db_rm_fileconflicts(const map<pair<pair<string, fileinfo_t>, conflict_t> > &files, const set<string> &keeplist);

Can you read this at first sight? Can you explain what it suppose to do? I'm not. I have just write this code a second ago and I can't parse it in my mind. I'm not going to waste my own time on crap-code, which I can't read properly. You have to change your preference or will read code like above.

Of course nobody forces me to follow your rules, I can simply leave this patch in my own collection (which already have much of crux-related patches). Have you ever thought that your attitude might be one of the reasons

Anton, Please keep the discussion on the list (I quoted your message fully to keep the others in the loop). My private e-mail address is no place for crux development questions. On Wed, Jul 12, 2006 at 01:52:39 +0400, Anton wrote: that your patches don't make it upstream?

But really I don't get your rules. pkgutils is so small package, and I don't understand why you're playing bombastic big-packages games. Are you saying that only big-packages should be changed incrementally, while small ones should get radical make overs only? I fail to see the logic of this, sorry.

And why I have to apply so many forces to be heard in this small crux-world? Because all I see from you is complaining (pkgutils-ng), whining (depot patch) and stating how you are right, and I am wrong (this thread). I have very little interest to work with such people, especially since many in the CRUX community manage to communicate in a friendly way, and manage to work WITH us as opposed to you, where I get the impression that all that matters is to have everything your way.

It took me quite some overcoming to fairly review your patch in the first place considering your former contributions to this list, and with every reply you sent in that thread it becomes more and more obvious that I again wasted the time even replying to a message from you (just like in the pkgutils-ng thread). Maybe you'll find someone else to merge your changes, otherwise just move it directly to your own collection. HAND, Johannes -- Johannes Winkelmann mailto:jw@smts.ch Zurich, Switzerland http://jw.smts.ch