| |
Scout
Registered: Dec 2002
Posts: 1568 |
Event id #2082 : X'2014
<Post edited by moderator on 14/3-2013 16:00>
Something to put in your agendas: X'2014 - October 24-26, 2014. |
|
... 301 posts hidden. Click here to view all posts....
|
| |
CreaMD
Registered: Dec 2001
Posts: 3035 |
Cool reading. |
| |
booker
Account closed
Registered: Jul 2003
Posts: 333 |
Quoting Axis/Oxyron(...) I´ve been there, done that, never will do it again! (...)
Why not, the demo is cool.
Quoting Axis/Oxyron
My advice: Dont burn these people. If they are gone, the scene is dead!
And that's very fair point, imho.
Actually, my advice (not that anyone would care): stop worrying about ppl saying group co-ops are bad - any reasoning invalid. Fuck that! |
| |
hedning
Registered: Mar 2009
Posts: 4601 |
Quote: Yep, its a bit like that. The critical point is the linking. I think at the moment there are 5 coders out there that have the experience to do proper linking for a competitive trackmo and have the will to spend the 500+ hours to do so, without getting any benefit or fame for it. I´ve been there, done that, never will do it again! So all other coders, artist and musicians crowd themselves around these 5 people. The good thing on the coops is, that atleast the knowlegde gets shared.
My advice: Dont burn these people. If they are gone, the scene is dead!
We do have Spindle 1.0 these days, but perhaps that isn't flexible enough? I am no coder, so I wouldn't know, but I do know lft is some kind of a genius. :) |
| |
cavey
Registered: Jul 2002
Posts: 68 |
C0: thanks for the feedback! Will device some tournament strategies :D
No complaints about the preliminary knockout phase? |
| |
Bitbreaker
Registered: Oct 2002
Posts: 500 |
Quote: We do have Spindle 1.0 these days, but perhaps that isn't flexible enough? I am no coder, so I wouldn't know, but I do know lft is some kind of a genius. :)
Had a look at it. Not applicable.
In our case we wasted the following for non effect stuff:
Global used ZP addresses: $02-$09
Reserved space for the loader and depacker (standalone + on the fly): $0200-$03ff
Linking framework for a bit less pain, to have a base irq and some save exiting load/depack calls that won't be overwritten on load/depack: $0800-$089x
Music: (if not 1 rasterline player) $0900-$1fff
Rest of the memory all grabbed by Bob.
Now look again at spindle that offloads several tasks to the c64 and by that also grows in size. Also the pack ratio of the included packer is not that overwhelming, this could mean even less parts per diskside. I really had my reasons for squeezing all into such a tight space with bitfire.
Things are fast though with spindle, no doubt, but what a demo has never enough of is loaderspeed _and_ mem. Furthermore people are forced to stick to the linking scheme given by spindle. It was already hard enough to give people enough possibilities to build the demo on their own, with 4 different supported assemblers, a Makefile and alternative autogenerated scripts to enable builds on linux, macos and windows for less skilled people.
And here, the sabotage hasn't even started, with parts that work fine standalone, but break when linked as they accidently destroy registers or the stack due to mistakes in the irq routines. Nothing where a framework can lighten your burden. I am willing to share the gained knowledge, but believe me, it is not explained with a simple hint to some trackmo system. We talk about a project here that had a repository with ~2800 commits in the end, 650Mb of data in there (okay, some checked in heavy tools, me included, haha)
Most of all, it is also about proper planning, and urging people to finish things in a certain order to bring linking forward and there need to be a subset of certain design rules put up beforehand, like the reserved memory areas.
Feel free to open a separate thread for all that :-) |
| |
Knut Clausen
Registered: Apr 2013
Posts: 18 |
@Axis @Bitbreaker - As you know, I had hoped for two separate releases by Censor and Oxyron. You make valid points that I haven't considered. GRG never complains about the linking of our demos, but it must be a terrible job. Thanks for enlightening me on this.
And nice meeting you guys at X btw. :) |
| |
Shadow
Account closed
Registered: Apr 2002
Posts: 355 |
This might be off-topic, but I think the discussion regarding trackmo-linking is a very interesting one.
To me it feels like this is perhaps one of the hardest parts when making a demo nowadays. I've only attempted it once, and it was a total nightmare, super boring work compared to writing effects.
As someone said, there is no "glory" in doing the work, but the general demoviewer expects that smooth, seamless ride with perfect transitions all the way.
I must confess, as a watcher of demos I also prefer smoothly linked trackmos, but as a coder, I would want the press-space trend to start again! :) |
| |
Radiant
Registered: Sep 2004
Posts: 639 |
Am I the only one who finds linking easy if done right? Sure, you have to adhere to some restrictions and a common format, but as long as you do that it's really not much - or hard - work to link a basic trackmo. Doing advanced stuff is trickier, but it's still often possible to generalize.
I find having a common framework (and assembler) helps a lot. Every coder involved in the demo should ideally be able to link everything together just by typing "make", as well as run individual parts in standalone mode. I think Lft's approach in Spindle is sound, but IME the same things can be achieved without imposing as many architectural decisions on the coders.
Manually linking together parts coded in a hodgepodge of different assemblers and with different memory usage conventions, buggy adherence to linking rules et cetera is however a nightmare, and whoever does that is both a hero and insane. :-) You save a lot of time by constantly integrating with the rest of the team instead of working independently. |
| |
Shadow
Account closed
Registered: Apr 2002
Posts: 355 |
In my trackmo I had coded all the parts myself, and I thought *that* was hard enough! :D
The actual linking is only part of the problem though, you need to have written all the parts with trackmo in mind from the beginning (ie. all tables/speedcode generated runtime to keep size and loadtime down etc.) |
| |
Radiant
Registered: Sep 2004
Posts: 639 |
(I'm probably a bit weird though, since I enjoy this kind of architectural work more than I do boring effect coding. :-P) |
| Previous - 1 | ... | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 - Next |
