Llamasoft does TxK for Sony Vita

[Markie]

We had some guests visiting but we also done stuff, maybe later on this week end some blog update

ooh, looking forward to that then :-) ... bring it on chaps :-)

[Markie]

We had some guests visiting but we also done stuff, maybe later on this week end some blog update

ooh, looking forward to that then :-) ... bring it on chaps :-)

nudge

[gilesgoat]

We had some guests visiting but we also done stuff, maybe later on this week end some blog update

ooh, looking forward to that then :-) ... bring it on chaps :-)

nudge

".. I managed to bufferize some stuff improving the rendering times a lot and now about to create a separate thread to
render some stuff in a different way .."

All I can say in 30 seconds now

[Yak]

We've had visitors, I've been mucking about in VR (and have an Oculus Rift ordered as a result), we fixed a gnarly bug, I've had to mow the grass, Giles has been working on UI and me on the main game, wanting to keep it true to its roots while adding new stuff. Giles' new explosion shader rips the web apart just perfectly. I have a class of particle called "plasmaball_particle". Made a weapon out of that.

Powerups going in, I need the powerup chain working so I can do the warp levels. I have a couple of transition warp sequences that just fit the theme perfectly.

I have a lot to do and i wish i could clone me an extra to work on sound design.

Patience.

[Jamesac]

Actually that was a great update, thanks.

James

[gilesgoat]

Actually that was a great update, thanks.

James

That gnarly bug was quite a bastard, one of those things that wasn't making the stuff crash but was making it go odd and
was taking between 30 and 40 minutes to appear ( sometime few minutes but that was a rare case ).

Already seeing what was going odd was deceiving because what you were seeing was not exactly what it was.

The first step been identifying what was actually happening.

The second step been trying to make so when it was happening it was actually stopping.

The third step been trying to analyze the stack frame and such to see where the program was at when this what happening.

This resulted in one of those cases "I suspect the stack history is not very accurate or something even stranger is going on", it was looking like "every time the crash was happening in a different part of the code" yet the cause of the crash
was looking always the same.

Then we tried the "divide and conquer" method of disabling/enabling bits of code to see if we could "bracket" where this
this cause was, this proven to be quite difficult as "at most" we managed to identify a relatively large chunk of code where
many things were happening and we could not go that much "in finer detail".

So what it looked like was like "some rogue pointer" blasting 8 bytes of memory and strangely enough always in the same
place, that gave us 'a bit of a hint' but not much , the other hint came out when we actually looked at what data was
put in and this was looking like some "meaningful" floating point numbers.

Then we tried using a 'data breakpoint' to actually try to stop excution when something was going to alter that location
this again proven to be not super useful, it proven the point causing the blast was some "internal function" but wasn't
giving much of a hint about what this really was and the function was not clear to identify it.

Then we noted the data in strangely looked almost always the same and after a better look of the numbers around we
got the idea that could have been a colour rather than a vertex coord.

At this point we made something to print and dump all the possible "set color" we were doing around and trying to
compute a difference with that suspected value until " == " and see if we could have found the thing that was causing it.

After a number of runs we findally found what was the "set color" that was causing the shit and luckily for us enough it
wasn't even a "dynamic" one it was one setting a certain colour.

Then it's been like "why this thing is causing that shit ? Is so innocent ?".

But 10 mins later or so we understood why.

Fundamentally "a strange and a bit rare combination of states and timings" and this function not within a certain "if"
a situation where one set of things were ending up and another set of things had to start but cause some randomised
values and other things once in a while there could have been an instant of time where "all was over and all was not began
yet" so actually in the middle of "for an instant X, there's absolutely nothing of this set to draw".

Probably that instant was one frame only.

Nothing of that kind to draw, but the setTint was there, trying to set a colour into a parameter area that was not existent
because nothing of that kind was selected but the function can not know it ( for performance reasons, the check would
significantly impact performance ).

And for a strange fate that thing was going to blast a small vertex buffer causing a curious effect and it was not
crashing anything because it was poking a "reasonable value" into an area that was designed to be written.

Fix : move that setTint inside the "if" it should have been.

End of the bug.

But it took quite a while to find all this out.

That's the bug story, it would be another long and me kinda-ripping-hair from my head story about the text-renderer
stuff and the ui stuff I am making

The explosion shader even another story, we knew what we wanted to do since start and I had an idea about how to
make it, I spent about 2 days and a half on it ( plus another 3 days to change some 'internal stuff' to make it easier to
work with that kind of stuff/create new shaders ) and then it was "kinda is there but I can't make it look exactly like
we had in mind".

Then I gave it to Yak anyway confident that 'with his magic touch he would have found the way', I was right.

From "kinda right yes it does but it's not yet 'that thing' " he changed a few little thingies and WOW, there it is !

"Jeff what is that cosine you put it there ? ... it does not make ( mathematically ) any sense .. "
"Yes I know, but it looks nice is it ?
.. yes .. it really does ... it really does it ... ... ( cosine ? ) .. (*)



(*) don't ask, it just works and looks nice ..

[Harlequin]

Would I be correct to assume that Yak can now speak Italian very fluently, as long as he only wants to swear?

[gilesgoat]

Would I be correct to assume that Yak can now speak Italian very fluently, as long as he only wants to swear?

Not super fluently but in terms of swearing he learnt quite well, it's quite fun to hear him swearing in Italian with his accent

[Daniel]

That gnarly bug was quite a bastard, one of those things that wasn't making the stuff crash but was making it go odd and
was taking between 30 and 40 minutes to appear ( sometime few minutes but that was a rare case ).

Already seeing what was going odd was deceiving because what you were seeing was not exactly what it was.

Its always frustrating hunting a bug like that, when you don't understand what is going on. But when you finally figure it out, you walk around with a smile on your face for the rest of the day. Well done. Give yourself a reward.

[gilesgoat]

Yes .. but just right now I introduced some new things that seems to work like a charm and that gave me
a smile this big

.. that did last about 15 mins while all in a sudden I started to have a NEW bug I never seen before ... but this time
the bug seems to be "lot less exoteric" and lot easier to happen, only that I am still trying to figure out now what is
exactly that is happening ...

Now I have a "heapalloc()" that returns NULL and of course that should not, I suspect there's something slowly munching
memory somewhere but I yet have to find where and what.

I have anyway a few ideas about what to try to do to try to see something.

[sdetoni]

Its always frustrating hunting a bug like that, when you don't understand what is going on. But when you finally figure it out, you walk around with a smile on your face for the rest of the day. Well done. Give yourself a reward.

Wish I could be happy about development at the moment. I was handed a project to finish written in Oracle ADF; if you don't know what it is, think yourself lucky! Its the most bloated, dis-functional, convoluted frame work I've ever seen. The idea is you write minimal code to build web page based applications, and link controls using bindings and some java backing bean concept.... but fails so bad, think configuration nightmare! The Web page designer is suppose to be WYSIWYG, but its not even close; page flow is design chaos; datasources don't refresh correctly, and hidden options all over the place that even the developer can't remember after a week away from the web page/code. And if your very unlucky, Jdeveloper will corrupt a webpage your working on preventing you from editing it ever again.... Its like Oracle Forms but 3 times worse!

Some of the more complex controls look interesting like, graphs, db table grids etc... but the pain to get there is so bad its just not worth the effort.

So I caned the ADF presentation, and rebuilt it in plain jsp, at least the source and page flow is now understandable, and while I have to code in HTML and CSS, at least I have better control on the outcome.

I usually don't get defeated by a language/framework but there even limits to my frustration that I can't take. I usually don't swear at work, but everyone heard me last week working on ADF! Cazzo che merda!

[gilesgoat]

Its always frustrating hunting a bug like that, when you don't understand what is going on. But when you finally figure it out, you walk around with a smile on your face for the rest of the day. Well done. Give yourself a reward.

Wish I could be happy about development at the moment. I was handed a project to finish written in Oracle ADF; if you don't know what it is, think yourself lucky! Its the most bloated, dis-functional, convoluted frame work I've ever seen. The idea is you write minimal code to build web page based applications, and link controls using bindings and some java backing bean concept.... but fails so bad, think configuration nightmare! The Web page designer is suppose to be WYSIWYG, but its not even close; page flow is design chaos; datasources don't refresh correctly, and hidden options all over the place that even the developer can't remember after a week away from the web page/code. And if your very unlucky, Jdeveloper will corrupt a webpage your working on preventing you from editing it ever again.... Its like Oracle Forms but 3 times worse!

Some of the more complex controls look interesting like, graphs, db table grids etc... but the pain to get there is so bad its just not worth the effort.

So I caned the ADF presentation, and rebuilt it in plain jsp, at least the source and page flow is now understandable, and while I have to code in HTML and CSS, at least I have better control on the outcome.

I usually don't get defeated by a language/framework but there even limits to my frustration that I can't take. I usually don't swear at work, but everyone heard me last week working on ADF! Cazzo che merda!

I can partially feel your pain, no thanks goat I don't know ( directly ) those kind of things but to my knowledge "there's not
ONE probably of those frameworks around that does really and so easily what it claims it can do".

I think part of the problem is that really "there's not easy way to connect A to B" I think I saw quite a few of those
frameworks having some "best case scenarios" but they can get incredibly bad/complex in what it's out of their "best case".

The thing I got "close" to such kind of things "recently" been when I done some stuff using the Android interface builder
that "just to give a look" yes ok you have an editor you drag and drop and put things together, KINDA, but when you have
to programmatically alter/create contents it's "fucking mess and not clear documentation/examples and too many troubles
with different APIs version".

In various things I've seen of various "frameworks" of other things I had my "general opinion/impression" is "I really do not
like frameworks or such", often because :

- yes they do the things but they do it in a very complex way
- sometime you get an airplane where all you wanted was a bycycle
- 'do as I say, don't do to much out of it' because if you do I can't work
- they can be "the best way to make 1001 things" but also "but none of which is much optimised, actually perfomance can be shit"
- their weakness resides in their strenght, to do so many different things, they are not very good in none

"The big final question" -> "How long is going to take me to learn to use and how this whole framework works and allow
me to do what I want to do the way the framework forces me to do VS how long is going for me ( and how much better )
to do it by scratch by studying the APIs" ?

So far, in almost ALL cases, the latter always beaten then first, for the kind of stuff I need to do.

In quite some cases also "yes but this framework does not really allow me to do what I want to do because it's not
really designed for what/how I want to do it".

Often I studied and understood bits and pieces of those things and used only the bits and pieces I needed by taking them
out and re-modifying them my own to fit the case.

[gilesgoat]

A quick one : I spent the last 4 days over a problem of "multi threading rendering" i.e. to have a thread that "prepares" some
stuff and when it's ready tells another thread "Ok now you can display this thing".

Turned out to be lot more tricky than initially considered once "a few other things" been factored in, it could have been
solved probably in a couple of other ways but did not want to use too PS Vita specific stuff so I used a "more portable"
approach instead.

One of those situations where the "main thing" of the code comes out like 30 lines of code or even less but it took me
4 days of various expriments, scribblings on paper, and various WTF ? WTF ?

There are some "peculiarities" of the Vita that make some sort of stuff a bit more complex to handle but allows you to
handle it in better ways (i.e. all the stuff is "there available" but you have to know/understand really well what you are doing
and what is going on when you do something ).

And yes, multi threading gives you headaches.

[gilesgoat]

Mini bit of extra info.

Yak is adding various stuff, we are a bit "tight schedule for this month" so we spend all time doing stuff on it, there's been
not much writing about what we are doing because we are too taken into making things work.

Basically we hit a bit of "performance issues" I mean yes it gets to a point where "if you ask a bit too much to certain things
they give what they give" so it's like "ok, how much to cut / how much we can do certain things in a differnt/better way".

But there's also the "balancing" between "how much I can spend on this VS things I have to do to deliver on time".

So far anyway so good all seems going quite according to plans, there's still lot to do anyway.

[gilesgoat]

I just been doing the "YES IT WORKS ! Happy dance" around the room cause after MUCH reading "notes of this", "notes of that", "instructions about this" "document number that", "refer to development notes appendix A paragraph .. " and such .. I finally manage to do a "PS Vita" package that works and installs TxK ( pre pre pre pre something ) on the Vita.

Which may seem "nothing" but to me it's been "something" as yesterday I was literally sinking into a sea of PDFs plus a little glitch I had during the SDK setup that
made me miss some stuff I had to re-download and such.

It's really a awesome feeling to see actually that shiny icon there you click it and the program runs .. especially when yesterday around 3 PM or so I was still with
hands in my hair with at least 8 PDFs open trying to understand "OK what is next that I need to do/know ? What the heck are they talking about ?" ..