PirosB3 · March 20, 2014 00:03 · PirosB3 · Mar 20, 2014
diff --git a/gistfile1.txt b/gistfile1.txt
 [10:57pm] PirosB3: I updated the doc
 [10:58pm] freakboy3742: I take it this is your daily "how's my proposal" conversation?
 [10:58pm] freakboy3742:
 [10:58pm] • FunkyBob makes a note that freakboy3742 is 3 hours behind
 [10:58pm] freakboy3742: FunkyBob: I am now. No DST in Perth.
 [10:58pm] FunkyBob: you lucky sods
 [10:58pm] PirosB3: freakboy3742: I admit having used the work component once
 [10:58pm] freakboy3742: Yup. Keep putting it to referendum, and it keeps losing.
 [10:58pm] PirosB3: I am guilty, but I think I'm using it the write way
 [10:59pm] freakboy3742: PirosB3: Heh…
 [10:59pm] PirosB3: ahah
 [10:59pm] freakboy3742: You're at a disadvantage - I came up through Comp Sci in the era of CORBA.
 [10:59pm] PirosB3: ahah
 [10:59pm] FunkyBob: hahahaha
 [10:59pm] freakboy3742: Most of the scars of that experience have healed… 
 [10:59pm] FunkyBob: I still remember dreaming of redesigning X as CORBA 
 [11:00pm] FunkyBob: how foolish and naive we were
 [11:00pm] freakboy3742: My co-founder actually used to be a senior manager at Iona.
 [11:00pm] FunkyBob: freakboy3742: don't suppose you have any experience with developing CFGs?  Alex is [understandably :)] ignoring my pleas 
 [11:00pm] PirosB3: I have never used CORBA
 [11:00pm] freakboy3742: FunkyBob: Only at the "you have to do this to pass CS" level. Haven't looked at them for a long time.
 [11:01pm] PirosB3: I don't know if that's a good thing or a bad thing
 [11:01pm] freakboy3742: PirosB3: Probably a good thing 
 [11:01pm] Siecje: FunkyBob: What are you trying to do?
 [11:01pm] freakboy3742: But, as Santayana says - those who forget the past are doomed to repeat it.
 [11:01pm] FunkyBob: Siecje: express tag syntax as a BNF so I can use rply
 [11:02pm] FunkyBob: CORBA... like OLE without the major flaws, but designed by committee
 [11:02pm] freakboy3742: That's the one! 
 [11:02pm] Siecje: FunkyBob: an issue in particular?
 [11:03pm] PirosB3: CFG = context free grammar?
 [11:03pm] Siecje: yeah
 [11:04pm] PirosB3: looks cool
 [11:04pm] PirosB3: lots of brackets
 [11:04pm] PirosB3: is it Lisp?
 [11:04pm] PirosB3: lol
 [11:05pm] FunkyBob: Siecje: I have no idea how to design one   I understand how to read them
 [11:06pm] Alex_Gaynor: FunkyBob: if you read the PLY tutorial, it describes a lot of the needed concepts
 [11:06pm] Alex_Gaynor: (sorry I've been in and out)
 [11:06pm] Siecje: PirosB3: You want a language to be context free so you don't need to know what is around a symbol to know what it is doing
 [11:07pm] dlogs left the chat room. (Remote host closed the connection)
 [11:07pm] FinalAngel left the chat room.
 [11:07pm] dlogs joined the chat room.
 [11:07pm] FunkyBob: Alex_Gaynor: yeah, it does... but... not quite enough of the "best practices" 
 [11:07pm] PirosB3: can you give me an example Siecje ?
 [11:07pm] Siecje: for example C is not context free because of the * operator
 [11:07pm] FunkyBob: PirosB3: https://www.google.com.au/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CCkQFjAA&url=http%3A%2F%2Fen.wikipedia.org%2Fwiki%2FBackus%25E2%2580%2593Naur_Form&ei=RyMqU5z_CJC3lQX8zoDwAw&usg=AFQjCNEqmyX1o21fYNzLCUP1yfOexI0KXQ&sig2=sRnfv2nz3Gi456jvmreerw&bvm=bv.62922401,d.dGI
 [11:08pm] FunkyBob: fucking google
 [11:08pm] FunkyBob: http://en.wikipedia.org/wiki/Backus%E2%80%93Naur_Form
 [11:08pm] jarshwah: dat url 
 [11:08pm] PirosB3: lol
 [11:08pm] PirosB3: aha!
 [11:08pm] Siecje: PirosB3: are you familiar with C?
 [11:09pm] PirosB3: more or less
 [11:09pm] Siecje: you are familiar with C++ because I read that in your proposal 
 [11:09pm] FunkyBob: Alex_Gaynor: nost of them use the very simple "basic math" example...
 [11:09pm] PirosB3: ahah 
 [11:09pm] Alex_Gaynor: totally.
 [11:09pm] PirosB3: yes, I am using it for uni
 [11:09pm] PirosB3: it's DirectX
 [11:09pm] Siecje: ?
 [11:10pm] FunkyBob: Alex_Gaynor: as an example... if the first token of an "expresion" is a NAME, I need to handle it differently [make it a context subscript]
 [11:10pm] FunkyBob: would i do that with a parser state flag?  or can I express it in the BNF?
 [11:10pm] PirosB3: for one of my assignments I am making a game in DirectX
 [11:10pm] Alex_Gaynor: Sorry, I'm not sure I understand what you're asking.
 [11:10pm] PirosB3: so obviously C++
 [11:11pm] PirosB3: I was answering Siecje
 [11:11pm] FunkyBob: Alex_Gaynor: that's ok... let me elaboate
 [11:11pm] FunkyBob: Alex_Gaynor: so, I'm writing a template parser... a sample might be :  {{ a.b[c].d["e"] }}
 [11:11pm] FunkyBob: which would translate to  context['a'].b[context['c']].e['e']
 [11:11pm] EvilDMP left the chat room. (Quit: EvilDMP)
 [11:12pm] Alex_Gaynor: you don't express the context stuff in the grammar, you express those in the actions
 [11:12pm] FunkyBob: plink
 [11:12pm] FunkyBob: context free grammar... of course
 [11:12pm] jarshwah: PirosB3: reading through your proposal now, just wanting to throw out some questions while I'm reading
 [11:12pm] FunkyBob: actions being the stuff decorated with productions?
 [11:12pm] PirosB3: nice! sure 
 [11:12pm] PirosB3: comment too!
 [11:12pm] Alex_Gaynor: yes, actions being the functions
 [11:13pm] FunkyBob: ok
 [11:13pm] PirosB3: in the meantime, I'm looking at this CFG 
 [11:13pm] jarshwah: I'm not sure I like the `with_model=True`.. wouldn't it be nicer to always return the with_model variant, and let callers disregard the model part of the tuple if they don't need it?
 [11:13pm] jarshwah: so get_fields() returns [ (field, model) ]
 [11:14pm] jarshwah: I'm assuming it returns a list of 2-tuples here
 [11:14pm] PirosB3: oh wow, 4 people on my doc. I'm popular
 [11:14pm] FunkyBob: Alex_Gaynor: with rply, productions which take non-terminals will be passed exactly what my other actions returned, yes?
 [11:14pm] Alex_Gaynor: yup
 [11:14pm] PirosB3: jarshwah: that is very true
 [11:15pm] FunkyBob: in this case, the're ast nodes...
 [11:15pm] PirosB3: jarshwah: I'll write that down
 [11:15pm] FunkyBob: am just trying to form in my head how the action will know it's the start of the expression...
 [11:15pm] jarshwah: I have similar thoughts on the local=False for get_concrete_fields, though I'm unsure what the difference is between local and non-local fields without looking into it too much
 [11:15pm] FunkyBob: but I guess if it's passed a NAME as the left, it knows it's the start...
 [11:15pm] jarshwah: but the API looks somewhat ugly to me
 [11:16pm] jarshwah: ..both the proposed one, and the existing one 
 [11:16pm] jarshwah: you've done lots of work here, so if what I'm saying doesn't really make sense, feel free to disregard
 [11:17pm] jarshwah: would you prefer me to comment on the doc rather than here on IRC?
 [11:17pm] FunkyBob: Alex_Gaynor: so, something like http://dpaste.com/1746492/
 [11:17pm] Alex_Gaynor: seems plausible
 [11:18pm] PirosB3: does anyone of you do machine learning by any chance?
 [11:19pm] PirosB3: jarshwah: oh sorry I didn't see you commented
 [11:19pm] PirosB3: yes of course! let's put it on the doc. So it's easier for other people to see the discussion
 [11:19pm] ambv left the chat room. (Read error: Connection reset by peer)
 [11:19pm] jarshwah: sure
 [11:19pm] FunkyBob: Alex_Gaynor: will write it up and see how it goes
 [11:19pm] mkai joined the chat room.
 [11:20pm] PirosB3: freakboy3742: Nice try  How do you know that 2 1 week iteration cycles will be enough?
 [11:20pm] PirosB3: ahaha
 [11:20pm] PirosB3: what do you suggest I put down?
 [11:21pm] bynaryshef joined the chat room.
 [11:26pm] FunkyBob: parser.py:1: ParserGeneratorWarning: 1 shift/reduce conflict
 [11:27pm] brts left the chat room. (Ping timeout: 240 seconds)
 [11:27pm] FunkyBob: Alex_Gaynor: http://dpaste.com/1746505/ comments welcome 
 [11:31pm] dlogs left the chat room. (Ping timeout: 265 seconds)
 [11:31pm] freakboy3742: PirosB3: Ok; that's review number 2 done.
 [11:31pm] freakboy3742: Better; still needs some work, especially in the timeline.
 [11:31pm] PirosB3: lol
 [11:31pm] freakboy3742: And an explicit description of the testbed sample project you intend to tackle.
 [11:32pm] PirosB3: ok with regards to the tickets
 [11:32pm] PirosB3: so with regards to the implementation and timeline
 [11:32pm] PirosB3: this is what I would like to do
 [11:32pm] PirosB3: my idea is a very "agile" approach
 [11:33pm] FunkyBob: Alex_Gaynor: whilst I can understand why the error messages may need to be terse... I don't know how to debug the problem
 [11:33pm] FunkyBob: and this is why I seek an expert to bother 
 [11:33pm] PirosB3: in terms of doing small steps, and if I need to fail, i'll fail early
 [11:33pm] PirosB3: but if I need to define a timeline for 3 months
 [11:33pm] freakboy3742: PirosB3: A good idea - *but*
 [11:33pm] PirosB3: this can be messy
 [11:34pm] freakboy3742: The problem is that saying "I'm going to spend 2 weeks" doesn't give us any confidence that the task will actually *take* 2 weeks.
 [11:34pm] freakboy3742: You say "Refactor Meta" is going to take 5 weeks total.
 [11:34pm] freakboy3742: Is that a reasonable estimatee?
 [11:34pm] freakboy3742: How big a job is it?
 [11:34pm] PirosB3: freakboy3742: the problem is that everything is based on our evaluation
 [11:34pm] freakboy3742: Yes and no.
 [11:34pm] PirosB3: how many endpoints is it going to have
 [11:34pm] PirosB3: ?
 [11:34pm] PirosB3: there are all questions that need to be answered, so for this reason
 [11:35pm] freakboy3742: You should be able to develop a "wet finger in the air" estimate based on what is there.
 [11:35pm] freakboy3742: We're not adding any new functionality.
 [11:35pm] PirosB3: I prefer having more time, and possibly finish before deadline
 [11:35pm] freakboy3742: Lets say there's 20 API entry points at present.
 [11:35pm] freakboy3742: We refactor, and get that down to 15.
 [11:35pm] PirosB3:  "wet finger in the air"  ahah
 [11:35pm] freakboy3742: You say "1 day per API entry point", and there's 3 weeks.
 [11:36pm] freakboy3742: (or 4, if you want to use the initial API count)
 [11:36pm] freakboy3742: At a granularity of 1 day.
 [11:36pm] freakboy3742: Now - the variables here - how may entry points are we talking about, and how long will each chunk take?
 [11:36pm] PirosB3: so you think I should not be speaking in days/weeks?
 [11:36pm] freakboy3742: Days/weeks is find a the macro level
 [11:37pm] freakboy3742: What I'm trying to get at is that we (the mentors evaluating the proposal) want to get a feel for how well you understand the problem.
 [11:37pm] FunkyBob: Alex_Gaynor: so, I fixed the mistakes in my decorator (replace term with expr) and now there's _two_ errors
 [11:37pm] freakboy3742: Providing a detailed estimate is one way to establish how well you understand the current landscape
 [11:37pm] freakboy3742: We want to you to provide evidence that backs up your claim that you can refactor Meta in 5 weeks.
 [11:38pm] PirosB3: sure
 [11:38pm] freakboy3742: Not just work back from "I have 12 weeks, and I need to spend half of them working on the refactor.
 [11:38pm] PirosB3: but there are many important factors to consider
 [11:38pm] PirosB3: such as the evaliuation
 [11:38pm] freakboy3742: What you've written so far reads like splitting up the available time, not building up from what work needs to be done.
 [11:38pm] PirosB3: the answers to our solutions
 [11:38pm] freakboy3742: Not sure what you mean by "the evaluation"
 [11:39pm] SmileyChris: "discovery phase", I guess
 [11:39pm] freakboy3742: Youv'e given a pretty good analysis of the sort of fixes you're going to be applying to meta.
 [11:39pm] PirosB3: I mean the part where we agree on an API
 [11:39pm] PirosB3: SmileyChris: yep  discovery phase
 [11:39pm] PirosB3: I always get the words wrong
 [11:39pm] freakboy3742: Ok - so that might be a reasonable block for a week or two at the start
 [11:39pm] freakboy3742: *but*
 [11:39pm] PirosB3: still, everything depends from that
 [11:40pm] freakboy3742: again, you need to validate that 1 week is enough.
 [11:40pm] freakboy3742: PirosB3: Ok - perhaps I'm not being clear.
 [11:40pm] freakboy3742: This doesn't need to be a 100% perfect, completely accurate estimate where we're going to hold your feet to the flames and withhold money if you don't meet your deadlines.
 [11:40pm] PirosB3: oh good 
 [11:40pm] freakboy3742: Last year, my student's proposal was off by about 4 weeks by the end of the program, and I gave him a glowing review.
 [11:41pm] SmileyChris: "hot glowing coals" review?
 [11:41pm] PirosB3: ahah
 [11:41pm] SmileyChris:
 [11:41pm] FunkyBob: waterboarding review?
 [11:41pm] • freakboy3742 files that one away for later motivational purposes 
 [11:41pm] PirosB3: great
 [11:41pm] freakboy3742: What we're looking for is evidence that *you* understand how complex the problem is.
 [11:41pm] freakboy3742: Not that you've said "I'm gonna work for 12 weeks, and then stop".
 [11:42pm] mkai left the chat room. (Quit: Leaving.)
 [11:42pm] PirosB3: I understand the complexity of the problem
 [11:42pm] freakboy3742: So - true, we don't know exactly what the final API will look like, and initial discussions will affect exactly how long work takes.
 [11:42pm] freakboy3742: Says you 
 [11:42pm] PirosB3: ahahah
 [11:42pm] PirosB3: obviously I don't know anything
 [11:42pm] PirosB3: I know enough to get starting
 [11:42pm] PirosB3: and I think that is fundamental
 [11:43pm] freakboy3742: Sure - but although we don't know the final form, we know *roughly* how big the problem is now.
 [11:43pm] freakboy3742: The functionality of Meta isn't going to massively increase as a result of your refactor
 [11:43pm] PirosB3: I would like to tell you what I would like to do in the next 2 weeks
 [11:43pm] PirosB3: but what I do after, depends on the evaluation of those 2 weeks
 [11:43pm] freakboy3742: So we know how many entry points there are right now - use that as a basis for your estimate.
 [11:43pm] PirosB3: but there is other complexity to consider
 [11:44pm] PirosB3: all the other parts of the system that use Meta
 [11:44pm] freakboy3742: "The current meta has 20 entry points. These may be refactored into a tighter subset, but assuming it takes 1 day per existing API, this phase of refactoring will take 4 weeks"
 [11:44pm] freakboy3742: That's all we're looking for.
 [11:44pm] PirosB3: oh really?
 [11:44pm] PirosB3: ok file
 [11:44pm] PirosB3: fine
 [11:44pm] PirosB3: that sounds ok
 [11:44pm] freakboy3742: Yes - an indication that you have looked at the problem enough to make an informed estimate
 [11:44pm] PirosB3: but still
 [11:44pm] freakboy3742: or rather, an estimate that isn't *just* informed by the 12 week completion deadline.
 [11:44pm] PirosB3: so what you want is proof of estimation
 [11:45pm] freakboy3742: Yes
 [11:45pm] PirosB3: ok
 [11:45pm] PirosB3: that makes perfect sense
 [11:45pm] PirosB3: just 1 thing
 [11:45pm] freakboy3742: Put it this way - if I said "Rewrite Django's template system" - estimate 5 weeks
 [11:45pm] freakboy3742: would you believe me?
 [11:45pm] freakboy3742: No.
 [11:45pm] PirosB3: I can do that up to a certain extent, but I cannot guarantee this in real life
 [11:45pm] freakboy3742: Oh - for sure.
 [11:45pm] PirosB3: ahaha
 [11:45pm] freakboy3742: Like I said, we're not going to hold you absolutely to this.
 [11:45pm] PirosB3: because I want to use a far more iterative approach
 [11:45pm] freakboy3742: It's going to be indicative.
 [11:46pm] freakboy3742: Ok - so factor that iteration into your schedule.
 [11:46pm] PirosB3: ok sure, so I will back up my estimates
 [11:46pm] mkai joined the chat room.
 [11:46pm] PirosB3: freakboy3742: how would you put it in?
 [11:46pm] freakboy3742: Excellent - that's all we need.
 [11:47pm] PirosB3: ok say what
 [11:47pm] freakboy3742: Pretty much exactly as I did 2 minutes back.
 [11:47pm] PirosB3: ok great
 [11:47pm] freakboy3742: It's fine to have a 4 week block - as long as it isn't just "4 WEEKS: DO STUFF"
 [11:47pm] PirosB3: also with regards to the project
 [11:47pm] freakboy3742: 4 weeks of 1 day tasks - that's fine.
 [11:47pm] PirosB3: ahhh ok that's better
 [11:48pm] PirosB3: Also, with regards to my project (testbed)
 [11:48pm] PirosB3: the way I would do this is just after the meta refactor
 [11:48pm] PirosB3: before the decoupling phase
 [11:48pm] PirosB3: I'm good at prototyping (I hope!)
 [11:49pm] PirosB3: so we could experiment before actually defining the spec
 [11:49pm] PirosB3: but how can I explain this?
 [11:49pm] freakboy3742: Well, describe what you're going to do, and why.
 [11:49pm] freakboy3742: The why is an important part
 [11:50pm] freakboy3742: otherwise a fresh set of eyes is going to ask "Why is Django funding an alternate implementation of Meta?"
 [11:50pm] PirosB3: my paragraph "benefits" will explain this
 [11:51pm] freakboy3742: So - "We have a cleaned up Meta. To prove the documentation is complete, it's helpful to have a clean room implementation of the API. As a proof of concept, we're going to build a model wrapper of XYZ, and develop that wrapper to the point it can be dropped into Admin/Modelforms. This will require additional documentation and code cleanup in Django itself"
 [11:51pm] freakboy3742: (elaborate into multiple paragraphs, but hopefully you get the idea)
 [11:52pm] PirosB3: but this is in the timeline?
 [11:52pm] PirosB3: I didn't expect it to be so tough to write a proposal
 [11:52pm] PirosB3: lol :=)
 [11:53pm] FunkyBob: it's like a bank loan... "prove to use you don't need this."
 [11:53pm] freakboy3742: PirosB3: Hey - there's $5k on the line here. We're not just gonna give it away 
 [11:53pm] FunkyBob: though in this case it's more "prove to us you won't make the mentor have to do anything" 
 [11:53pm] freakboy3742: And seriously - you're getting away easy here.
 [11:53pm] FunkyBob: $5k?  where do I sign up??
 [11:53pm] freakboy3742: FunkyBob: Well, head down to RMIT and enrol 
 [11:53pm] FunkyBob: haha
 [11:53pm] PirosB3: ahahahaha
 [11:53pm] FunkyBob: Monash is closer
 [11:54pm] FunkyBob: what if I get my wife to submit the application?  
 [11:54pm] FunkyBob: does the mentor have to be a core dev?
 [11:54pm] PirosB3: freakboy3742: (potentially=) how much would we be communicating?
 [11:54pm] jarshwah: 1 term costs about $5000 these days 
 [11:54pm] freakboy3742: PirosB3: Well, as much as necessary 
 [11:54pm] PirosB3: daily?
 [11:55pm] freakboy3742: As an indication; last year, I had a weekly phone call that lasted an hour or so, and feedback on mailing lists during the week
 [11:55pm] freakboy3742: But that's not set in stone
 [11:55pm] PirosB3: that sounds good
 [11:55pm] freakboy3742: Essentially, you're expected to work independently as much as possible
 [11:55pm] freakboy3742: but the mentor is there to give you as much feedback as necessary.
 [11:56pm] PirosB3: freakboy3742: I can always contact you at your 7am 
 [11:56pm] freakboy3742: At the start, that might mean a lot more interaction, possibly less t the end.
 [11:56pm] PirosB3: I'm sure you will be delighted
 [11:56pm] freakboy3742: PirosB3: Oh goody! 
 [11:57pm] PirosB3: ok
 [11:57pm] freakboy3742: Assuming you get picked, and I'm the mentor (and I like your chances of both of those), we can shake out details as we get closer.
 [11:57pm] PirosB3: ahaha
 [11:57pm] PirosB3: cool
 [11:57pm] PirosB3: so now I have to convince you 
 [11:57pm] freakboy3742: Indeed 
 [11:57pm] PirosB3: so… in order to convince you
 [11:58pm] freakboy3742: Unmarked, nonsequential bills help. 
 [11:58pm] PirosB3: I will update the timeline, giving more detailed estimates. with a WHY
 [11:58pm] PirosB3: ahahaha
 [11:58pm] PirosB3: I'll send a carrier pidgin
 [11:58pm] freakboy3742: PirosB3: yes - and more detail on the testbed project (why and what)
 [11:58pm] PirosB3: great
 [11:59pm] PirosB3: then=
 [11:59pm] PirosB3: ?
 [11:59pm] PirosB3: also, how do I submit this?
 [11:59pm] PirosB3: they have a weird editor
 [11:59pm] freakboy3742: Yeah. that they do. Melange is… interesting.
 [11:59pm] PirosB3: is there now way I can provide a link?
 [11:59pm] freakboy3742: Essentially, no.
 [11:59pm] PirosB3: lol
 [11:59pm] freakboy3742: I've got a feeling this is part 1 of Google's plan - you must be able to work out Melange before you get our money.
 [12:00am] freakboy3742: It really is a clusterf*ck.
 [12:00am] PirosB3: basically I need to plan a day upfront to port my docs to Melange
 [12:00am] PirosB3: can I already submit, and then replace?
 [12:00am] • FunkyBob thinks he needs a neural tap into Alex_Gaynor's brain
 [12:01am] PirosB3: how do you do that FunkyBob?
 [12:01am] FunkyBob: no idea
 [12:01am] PirosB3: ahah
 [12:01am] PirosB3: third person
 [12:01am] PirosB3: PirosB3 says blablabla