fredpointzero @fredpointzero

Graphics and engine software engineer @ Unity. Paris, France Joined February 2008

Tweets

414
Followers

143
Following

338
Likes

39

Zara Alfonso Illustration @artofzara

4 years ago

Today is my birthday, I'd love if people would RT my art 🥰🖤

257 12K 33K 0 835

View Details

Jānis Lācis @Lidotajs

4 years ago

😬 #DontLookUp in real life. 🥵

621 22K 87K 0 5K

View Details

fredpointzero @fredpointzero

4 years ago

@Gazedsoul Very nice and detailed concepts! How much time have you spent to create one of these?

0 0 0 0 0

View Details

fredpointzero @fredpointzero

6 years ago

@richgel999 And efficiency is relative. Hiring a lot of devs to patch bad code quality, vs fewer engineer but better code quality has a big difference on costs for the same results

0 0 0 0 0

View Details

fredpointzero @fredpointzero

6 years ago

@richgel999 Also, the life of devs is way less stressful. The difference between « it will works » and « it will probably work »

0 0 0 0 0

View Details

@richgel999 Game engine should have a lot of clean code (at least a big part of it) It just makes cheaper and easier to add or maintain new features. And this is the only way to reach top software quality. But i feel this is not a concern, and that some lead can’t/won’t produce clean code

0 0 0 0 0

View Details

fredpointzero @fredpointzero

7 years ago

@Nifflas With user tests. And based on the knowledge acquired from the fast bad code phase, you can actually try to design and implement a reliable product if you intend to reuse it a lot. Fast bad code -> learn and throw away code -> design code adapted to your need based on learnings

0 0 0 0 0

View Details

fredpointzero @fredpointzero

7 years ago

@Nifflas But the cost of fixing bugs, or slow and hard authoring data, or misconception building on top of bad code. This is expensive and what should be avoided. But, writing fast bad code in order to learn is good. You can learn to have better skill, learn a feature’s worth

0 0 0 0 0

View Details

fredpointzero @fredpointzero

7 years ago

@Nifflas You can see it differently, writing bad code is fast and hard to pile up, writing good code produce reliable software and you can reuse it for other feature, but are way more expensive to produce. That being said, what’s actually bad is not the code itself

1 0 0 0 0

View Details

Henri Maler @HMaler

7 years ago

"Une version édifiante de la défense des libertés". A faire connaître largement ! (via @nicolasgregoire )

9 469 332 0 13

View Details

fredpointzero @fredpointzero

7 years ago

@Jonathan_Blow Some companies run viable business by shipping flawed software, they usually hire a lot of developers to implement a lot of features without reliability or quality in mind. This is also a model that has been effective and understandable for non technical, managing people

0 0 1 0 0

View Details

Donald Mitchell @DonaldPMitchell

7 years ago

"With the advent of the high speed digital computer, it will now be possible to make mistakes at a rate never before dreamed of." -- quote in the front page of the CDC 6600 FORTRAN manual from the 1970s.

1 67 200 0 3

View Details

fredpointzero @fredpointzero

7 years ago

@McCloudStrife There are actually methods to write software like mathematical proofs, it uses proven compiler on proven hardware. It is not bug free but it is the closest I am aware of. This is used to build critical software. Like Paris’ automated metro line 14, like 4 bugs found in 10 years

0 0 0 0 0

View Details

fredpointzero @fredpointzero

7 years ago

@sleepyGrinn @ErbGameArt Nice! But I think the shapes are still too perfect, raw intense energy is brutal and irregular, maybe some noises on the spirals

0 0 1 0 0

View Details

fredpointzero @fredpointzero

7 years ago

@Atrix256 Then when the final use case is covered, I stop to iterate. If there is another use case coming, then I resume the iterations.

0 0 0 0 0

View Details

fredpointzero @fredpointzero

7 years ago

@Atrix256 It depends on the stage of the code, prototyping: breadth first, API shows when error can occurs, but most parts are todos. Then I iterate several times and at each iteration I find which parts are valuable immediatly and implement those depth first.