Present: Tobi, zerver, abma, jK
<abma>hi all!
<abma>i'll make minutes
Release plan
<abma>who will make the release? jk again?
<abma>if so, whats left?
<jK>I am doing so right atm
<abma>aaah, nice
<abma>afaik current release branch is fine
<abma>did a testgame with zydox and it seemed to work fine...
<jK>nice to hear
conclusion: hopefully tonight a 87.0 release
"Map creation" in forum
<abma>http://springrts.com/phpbb/viewforum.php?f=13
<abma>as it currently is, its more a forum for releases
<jK>+1
<abma>my suggestion is: lock the tutorials subforum, because all tutorials seems to be to old and they should go to the wiki...
<abma>+ add a "map release subforum"
<abma>i can't do that, i've no access... :-/
<abma>[RoX]Tobi: can you do that?
<abma>i just want to keep the meeting at high speed
conclusion: lock the "Map Tutorials & Resources" subforum + link to http://springrts.com/wiki/Map_development
Google Summer of Code (GSoC)
<abma>i try to register at least, maybe we are lucky... (i know, having a gsoc will make much work)
<abma>but i would be happy if we could get / do it
<abma>so... if you have ideas, please add to http://springrts.com/wiki/Project_proposals
<abma>i tried to make it a more general page and not so gsoc related
<abma>(+ year independant)
<jK>really nice what you are doing there
<abma>thanks...
<zerver>yeah
<Tobi>+1
<abma>
<abma>i skip the points "merge luasocket / remove lualobby" ... because release + setting live comes first
conclusion: Google, please choose our GSoC projects, we'll attract the best coders!
possible solution for in-source lua documentation
<abma>i was a bit anoyed by the current lua-docs...
<abma>when i search for a function i need, i currently grep the source code...
<jK>nearly impossible cause of the way lua works
<abma>a possible solution would be, do the comments in the c++ source
<jK>-> one function & multiple syntaxes
<abma>and we use a custom syntax which is then used by a doxygen filter that prepares it for doxygen
<abma>i didn't fully investigate if it works as we need it
<jK>doxygen doesn't support multi syntax
<abma>if not, adding some output stuff for the script shouldn't be so hard
<abma>multi syntax?
<jK>Spring.PlaySound("foobar.wav") Spring.PlaySound("foobar.wav", volume) Spring.PlaySound("foobar.wav", pos_x,pos_y,pos_z, volume)
<Tobi>one C++ function mapping to multiple ways to call it in Lua isn't really a problem (although doxygen alone would be insufficient)
<jK>that's a bit more than simple overloading
<Tobi>harder problem may be those APIs that partially rely on Lua functions, in addition to the C++ stuff
<zerver>I agree it would be neat to have in-source docs, it is so easy to forget to update external docs
<jK>atm I would concentrate on auto-documentation of Unit/Weapon/CEG-defs
<jK>those are much easier to parse
<abma>hmm, i thought that a custom doc-string is added, at a place where it fits best
<abma>and i see, its much more complicated than i thought :-/ damnit
<abma>my idea was, to add the comment, directly to the function, for example for Spring.Echo, exactly here: https://github.com/spring/spring/blob/d ... s.cpp#L722
<zerver>Need some custom parser to extract these docs maybe
<abma>yeah, sure
<Tobi>extracting comments can be an awk one liner
<abma>doxygen seems to be a bit inflexible there...
<jK>Spring.Echo is already a special case but still easy to document with doxygen syntax (it uses vararg similar input)
<Tobi>only when you want to integrate it with knowledge of the actual code you face a learning curve with near infinite slope
<jK>the syntax variations of many other lua function are _not possible_ in c++
<jK>e.g. Spring.PlaySound("foobar.wav", volume) Spring.PlaySound("foobar.wav", pos_x,pos_y,pos_z, volume)
<abma>yeah, thats what i expected, but i don't know all cases
<jK>volume is a float & pos_x too
<Tobi>just don't use doxygen
<jK>so it is no c++ overloading
<Tobi>or at best, re-use the formatting a little bit, so the comments still look acceptable in doxygen
<abma>[RoX]Tobi: yep, that seems to be the only possibility as a doxygen filter isn't enough...
<Tobi>but implement custom extracter -> parser of the comment syntax -> generater of HTML
<Tobi>I wanted to do exactly this quite a while back, but was too ambitious; I wanted to also make the comments edittable on the web and automatically generate commits from the edits
<Tobi>combined with lack of time it didn't work out
<abma>[RoX]Tobi: hmm, with github, it is close to that...
<jK>documenting a function like Spring.PlaySound in a doxygen similar way would need >2 pages
<Tobi>abma_irc: yeah, github obsoletes this idea anyway (IIRC it was before github had web editor)
<jK>you would need a totally new syntax to eff. document those functions (e.g. each possible syntax combination is on it's own _single_ line, and the param themselves are document beneath separate)
<Tobi>jK: [21:56:52] <[RoX]Tobi> just don't use doxygen
<abma>yeah, thats maybe a conclusion, no doxygen
<abma>ok, thanks. thats many input.. i try to write it down to the project proposals wiki page
conclusion: in-source lua-api documentation will be a huge project: doxygen can't be used so it will be something like a tiny doxygen
status of touchdev branch
<abma>(i always forgot the name of the guy that wrote it, sorry)
<Tobi>the only thing that may be nice is if the comment doesn't look totally obfuscated in doxygen, but that could be a matter of wrapping the whole block in <code></code> or whatever the doxygen equivalent is for that
<abma>[RoX]Tobi: i would suggest /*- .... */ doxygen should ignore it completely then
<abma>ok bill wrote the touchdev stuff
<abma>he said, it seems to work fine, i also added some of his lua widgets that use it, but they have to be cleaned up first
<Tobi>that would work too
<jK>abma, check my todo list for things that should be done before merging it
<abma>[LCC]jK: thanks, will look into these
<abma>ah, bill also told me, he still works on it, but he has very few time
<abma>thats it for that point...
<jK>also TUIO should be replaced with XInput2.x
<jK>but that's a feature request
<abma>ok..
conclusion: the touchdev branch is alive, but there are still some points left to get it merged (mainly code-cleanup)
delete old git branches (those 5 0.82 ones + syncdebug-gcc-4.3)?
<jK>anyone needs them?
<jK>are there any important commits in those, not commited to dev-branch?
<abma>if still not, bad luck i would say...
<Tobi>can delete them afaics
<jK>k will do so then
conclusion: jk will delete those branches
extend Spring.Echo() to allow a loglevel parameter?
<abma>i need a echo function that allows to set the loglevel
<abma>i would use it for printing lua error messages to stderr
<abma>do you see a way to extend Spring.Echo() that a errorlevel can be set?
<abma>currently it accepts all as parameters...
<abma>my idea was, to check, if the last parameter is an int, if so, use that as loglevel
<abma>this should break very few lua code
<zerver>should work
<abma>if not, i would add Spring.Log()
<abma>with msg, loglevel as parameters, where loglevel is optional
<jK>bad idea
<jK>that one -> <abma> my idea was, to check, if the last parameter is an int, if so, use that as loglevel
<Tobi>yeah
<jK>if at all you can use custom lua userdata or strings
<zerver>Maybe first param instead?
<zerver>But Log may be cleaner...
<jK>so Spring.Echo("LOG_ERROR", "foobar") or Spring.Echo(LOG_ERROR, "foobar")
<jK>but I would prefer adding Spring.Log
<abma>yeah, this will break no existing code...
<zerver>Neither will Spring.Echo(LOG_ERROR, "foobar") ?
<zerver>you cant echo an int iirc, has to be string
<abma>hm, afaik Spring.Echo(1,2,3,"test","test") works...
<abma>Echo calls tostring()
<zerver>aha
<zerver>then definitely Log()
<abma>ok, thx
<abma>then we are finished :-D
<zerver>But I am sure I tried to echo something, and it failed. Maybe it was a bool.
<zerver>or nil
<jK>Spring.Echo() doesn't output anything
<Tobi>it could work if LOG_ERROR is a userdatum, but Spring.Log is cleaner IMO
<jK>same as Spring.Echo(nil)
<abma>https://github.com/spring/spring/blob/d ... s.cpp#L776 ?
<jK>but Spring.Echo(nil, 1) will print "nil, 1"
<jK>and yeah, in my example LOG_ERROR is not an int, it's a userdata
<zerver>yeah
<abma>hmm, Spring.Log seems to be much cleaner for me... even if its the same function with some tiny difference
<Tobi>yes, it *is* much cleaner
<abma>zerver: should your last commit go into release or not?
<Tobi>abma_irc: did you figure out about the mailing lists?
<abma>it seemed to work?
<zerver>it was an accidental float instead of bool, but IIRC it makes no difference
<Tobi>yeah
<zerver>true = 1.0f and false = 0.0f
<abma>[RoX]Tobi: then its fine...
<Tobi>ok
<abma>where do i define LOG_ERROR / LOG_INFO /... for usage in lua?
<abma>is there already some place for such consts?
<abma>or don't define and use 1, 2, ...?
<jK>hmm don't think it's already done in our lua api that way
<jK>will see if I can write it (it's not that hard, but you need some knowledge about lua's userdata types)
<jK>-> here `light userdata` is needed
<abma>aah, ok
<abma>+ thanks
<jK>np
<abma>thats what i've already: https://github.com/abma/spring/commit/c ... 3dc4e9daea
conclusion: Spring.Log() will be added
Does not work.