Erok needs developers!

[sli]

As many of you know, I've been working on a Python port of OpenKore by myself for some time. I've always viewed it as a companion project to OpenKore, not a competing bot (think WINE and ReactOS, not OpenOffice.org and Microsoft Office). The problem is, however, that this is a bit of a task for just one person. So, this is an open call to all Python developers. I have strict standards for those who wish to join the project, but nothing too bad.

Applicants must have a not only a functional understanding of the Python language, but also a firm grasp of advanced Python topics and be able to display their skills. Knowledge of OpenKore is not required but certainly preferred. I will not be doing any Python training specifically for this project, but I will most certainly school all applicants on the caveats of the porting process. I'm also looking for Python developers that are experienced with extending Python through C/C++ modules.

I don't think there are a lot of Python developers in this community, but everyone wishing to be a part of a project will be considered (read: ask your friends).

[Cozzie]

teach me python plz.

[sli]

k

[isieo]

ahem, perl developers wanted for openkore !!!

[sli]

NO HIJACKING MY THREAD! Besides, Kore already has more than one developer. Erok only has me.

[Failure]

teach me python too plz :]

[VCL]

Wow, when you said "port" you weren't kidding! I remember that half a year ago you didn't have any working code in your repository, but at this moment your code seems to mirror OpenKore's, just in Python syntax. Well I suppose that porting it to Python syntax, as opposed to rewriting it in Python, isn't so bad. Previous bot rewriting attempts have mostly failed because the job proved to be too much work, and the authors quickly gave up. If you want to make things work as soon as possible, with a feature set that's near OpenKore's current one, then porting might not be such a bad idea.

If I were to rewrite OpenKore I'd rethink its architecture completely. The current architecture has a lot of cruft and is the result of too much hack work, as well as the result of RO itself having a lot of cruft (at the protocol level). Kura's old code has a lot of Perlisms (of the unreadable kind) that I've been trying to get rid of for several years, though it was and still is a lot of work to refactor all that stuff.

Oh and I would have done it in Ruby, not Python. Using Ruby should also give scalability benefits compared to Python. I don't see Python getting rid of its global interpreter lock any time soon. Ruby on the other hand has JRuby which can fully leverage multiple CPU cores in a single process. Plus it's just a nicer language in my opinion.

[sli]

I've done a lot of... "Pythonic" changes in Erok already. Considering where it's gotten in less than 2000 lines (literally everything except for the AI and network code), I'm not going to doubt that Erok will end up being serious competition for OpenKore. I'm leveraging Python's strengths as much as possible, but in these early stages I'm trying to simply complete a feature-for-feature port, and then refactoring on a small scale by just optimizing bottlenecks with better code, something I'm used to doing. Right now the overall goal is to create a feature-for-feature rewrite in Python using only standard library modules as much as possible and circumventing the need for custom modules written in C. Right now the only code implemented in C is path finding, and I used Cython, Python's answer to XS, for that.

I'm glad that you seem interested in Erok, VCL (everyone else seems threatened by it ). Skimming a couple of your other recent posts, it seems like you and I would have a lot to talk about.

[VCL]

Well right now I'm not very involved with OpenKore. I can understand why others might feel threatened, and I admit that I myself am not totally feeling unthreatened - it was "my" project after all. But my absence allows me to view the situation from a more neutral stand point.

I have no doubts that Perl wasn't and isn't the best language to write OpenKore in. The reason why it's in Perl is because of the legacy that we've all inherited from Kura, with all the associated pros and cons. You say that Erok can become serious competition to OpenKore. Technically, any bot can become serious competition, but technology isn't the main issue here; manpower and dedication, are. OpenKore isn't exactly a piece of technical wonder (I tried, but there's too much legacy code and too many features that prevented me from doing things properly), but the reason why it's so successful* is because people continued to maintain it. The code may not be nice, but at least it's continually updated. Other attempts at rewriting OpenKore or at making a new bot that is technically sound, have failed because the authors gave up. Man power and dedication and continuous updates turned out to be more valuable to the project's survival than pure code quality.

So make no mistake. You seem to be proud of your work, and I can understand that. But also understand that you probably won't "win" if you rely too much on code quality as the only drive.

Oh and check IRC, I'm in your channel right now.

* In this post, I define "success" as "used by pretty much everybody in the RO botting community".

[h4rry84]

someone is Alive