Perfect vs. Available Parts

[Michael Heidemann]

This morning I looked through the files that are on hold on the PT.

My personal opinion on releasing parts is - good enough! - because of the following reasons.
1) Very seldom a part is visible in a picture in a close-up. So tiny errors are not visible.
2) If the parts are on the PT they are used - so they can't be that bad.
3) If I need a part and it is not in the library I am disappointed. If that occure too often i lost interest.

What is good enough!
1) Connectivity portions needs to be on the right place.
2) The origin is placed according our agreements.
3) The parts shape follows the visual appearance of the real part.

What should not be a reason not to cert?
1) T-Junction on flat surfaces.
2) Overlapping flat surfaces on opaque parts
3) Missing of some condlines that do not affect the overall appearance.

So do not wonder if I cert a part where you say - so many errors in the file.

I hope lot of you can follow my thoughts and on second iteration on the PT we can make the parts perfect.

[Philippe Hurbain]

I heartfully follow you!!!

[Magnus Forsberg]

Hi Mike,

To me there are no such thing as "good enough".

Let's start correcting the parts at the Part Tracker, instead of certifying them only because there old or "almost" correct.
We will only be caught forever correcting official parts, instead of creating all the new part LEGO produce.
I see no point in letting parts go, if they're only going to have to come back for later corrections.
Sure, it's easier today, to make better designs, with all the tools we have. Let's use those tools and data we already have,
and put all the personal pride aside, and start working on the designs we have.

I started to correct all the parts that I had Hold-voted early december last year.
Almost all were later cert-voted and released in the latest update. Hopefully without having to come back for corrections.

I don't wanna spend my time correcting other authors "misstakes" and "errors". I wanna make my own. ;-)
It's only after correcting your own misstakes, you'll learn to do things right.

[Steffen]

In the beginning I also was thinking like you.
I was striving for extreme perfectionism.
This led to the effect that some parts took very, very long to leave the PT the first time.
This meant that many users couldn't use them for building.
Nowadays, I've modified this thought pattern a little (we all get older, right...?):
I nowadays think that a 2-iteration principle is probably the most productive here:
1. get a part out as early as possible, with correct origin, usage logic (connectivity etc.) and BFCing
2. strive for perfectionism in a 2nd iteration on the PT
This brings a part out as early as possible and reduces the pain of waiting.
Additionally, it allows for taking all time to make the part really perfect.

I understand that not all people follow this principle. We're all different. So maybe we can find the minimal compromise of:
- if a part is not perfect and you don't wanna cert it: do not cert it, but also do not hold it
- use hold votes for severe geometry, logic or origin issues

This is just to share what's inside my head.

[Michael Heidemann]

1+

[Michael Heidemann]

I know

[Michael Heidemann]

I like your perfectionism! But not in the first step

There are so many issues that are really necessary to solve.

If you like to go your way - ok, but do not vote hold if it is really a minor issue. Thanks.
If it comes to parts i have created - please feel free to correct before hold

[Magnus Forsberg]

Ok, I won't hold vote your parts,
but then, please tell me why I should bother to review your files, if I already know that you don't want to correct them.

[Michael Heidemann]

That is not true.
If there are big issues I very much like to know that so it can be corrected - by me or someone else!!
It is difficult to draw the line between good enough or not.
I am not against perfect parts!!
I am against holding all parts until they are perfect! - By the way - what is perfect? We can only judge on the todays rules and commitments. Tomorrow they might change - maybe only slightly - but then the parts isn't perfect anymore. Therefore I say: Good enough for release does not mean perfect for today.

Let us work together for a good library, quality _and_ quantitiy

[Michael Horvath]

LOL

[Magnus Forsberg]

By all means, then let's slap a Needs Work on all the unfinnished parts and clean out the PT.

No, seriously, let's work with the files we have, so that they don't need to come back.
Yes, better tools and development of our standards will perhaps force them back to the PT,
but unfinnished designs should not be certified.

[Philippe Hurbain]

A part that is geometrically correct and looks good... is good!

[Michael Heidemann]

Any useful comment from your side ?

[Michael Heidemann]

Please do not go this way.

For example:
http://www.ldraw.org/cgi-bin/ptdetail.cg...s/6058.dat

This file is surely used in some models out there. It does it's job.
For sure I would _not_ have certified that in this condition, but it works for years.

Now, as we have it again on the PT we should release it after _all_ issues are solved.

Please think about this for a while. (If I want to go to a far away city, i do not wait for a Bentley, I use the car that can go that distance).

[Tim Gould]

@Michael Horvath> If you want a post of yours deleted, please do not mark it as spam. That's abuse of the spam reporting system.

Also, we rarely delete posts without very good reason. But if you really feel something should be gone, make a post titled "please delete my post above" (or words to that effect) and give a reason.

Tim

See http://forums.ldraw.org/showthread.php?t...4#pid11654

[Tim Gould]

This has been my feeling for a long time. I think we've got closer to it. I see parts going through that used to be held for minor mistakes.

Tim

[Tim Gould]

Hi Magnus,

The basic rule of thumb for coding and similar activities states that you get 90% of the way to perfection in 50% of the time. Certinaly this applies to part authoring too. So we can have two 90% parts in the same time it takes to make one 100% part.

By way of extreme example, if we really wanted the library to be perfect we would have to go through every single technic brick (and many other parts) and move the connection hole one LDU higher. Because there are some parts that simply cannot be made perfect while having the hole in the wrong place e.g. 98286. This is a massive waste of time that could be spent making new parts.

And I see you yourself have even worked around the problem in 53533. Which suggests that even a strong desire for perfection can be tempered in the face of excess time-wasting

Tim

[Michael Horvath]

Which Michael?

[Michael Horvath]

I prefer things the way they are now. Yes it's frustrating when a part you really want to use is not certified. But in those cases you can create an MPD or inlined part to distribute with your work.

[Tim Gould]

You. If both Michael's reported spam I'd be starting a new thread

Tim

[Tim Gould]

That's fine for people who are semi-expert at LDraw. But the current system can force people to become experts when all they really want to do is build using the latest parts. And that's not cool.

Tim

[Michael Horvath]

I don't recall reporting any spam. At least not on purpose. Are you absolutely sure it was me and not someone else?

[Ben Supnik]

Hi Michael,

What kinds of errors are you saying are both:
- Currently stopping reviewers from certing a file and
- Actually not so bad, so that we should cert and then fix it later?

Cheers
Ben

[Tim Gould]

Oh crap. My mistake. No it wasn't you. Sorry for the confusion and false accusation.

Your post was reported and I missed that the person reporting it was Max Martin Richter.

Tim

[Michael Horvath]

That's OK. Play well!

[Michael Heidemann]

I do not quite sure I understand your post.

But yes, both can happen.
1) I can not force someone to vote cert on a part. I can only convience them .
2) exactly this way, if _no_ _basic_ mistake is made in the file.

cu
Mike