Hide Idle (>14 d.) Chans


← 2019-07-16 | 2019-07-18 →
diana_coman: shrysr: are you coming in or going out ? this join/part is only annoying
diana_coman: oh well, apparently not that alive really
diana_coman: shrysr: are you there?
shrysr: yes
shrysr: Hi !
diana_coman: oh hi, so you're alive after all, lol
diana_coman: what brings you here?
shrysr: I want to do work that matters :)
diana_coman: heh, hold onto that idea! what do you know how to do?
shrysr: Sorry, I have been trying to setup emac and erc on my VPS and it always feels new everytime
diana_coman: well, vps is a sort of "too cheap to afford all the problems it brings" but anyway
shrysr: Not much! I do not know Ada. I've been building skills in R... I am familiar with python. I did know C / C++ a LONG time ago.
diana_coman: R is not bad to know, actually; still quite my go-to thing for messing about with data
shrysr: Yes, my current goal is to b
shrysr: 'break into' data science.
diana_coman: oh, what would "data science" be exactly, you think?
shrysr: I think it is about gaining insights from data, understanding it's nature and then plugging all of that into algorithms to extend that understanding further.
shrysr: I'm fascinated by the idea that a machine can learn.
diana_coman: well, insights won't ever come "just from data" but anyway, have you seen then for instance http://trilema.com/2016/the-pop-of-the-day-and-other-items-of-virtual-economy/ ?
diana_coman: there's plenty more in fact - in this sense, a working Eulora is effectively the data scientist's wet dream
asciilifeform: imho most 'machine learns' reduce to variant of http://www.loper-os.org/bad-at-entropy/manmach.html
asciilifeform: massively overhyped concept.
diana_coman: asciilifeform: like all fashions, it has to be hyped, what, don't begrudge it, its false feathers and fancy garb
diana_coman: shrysr: that fascination is most likely to be linked to what you add on "learn" than to what the machine actually does
asciilifeform: diana_coman: problem typically comes for people who go and try jump from cliff with the false feather wings.
shrysr: asciilifeform: I understand that. However, it still seems to have a lot more potential than mechanical engineering :D
diana_coman: asciilifeform: how's a problem if they jump? good instruction for onlookers, I'd say, let them JUMP!
diana_coman: shrysr: what are you working on/doing otherwise atm?
asciilifeform: diana_coman: the onlookers never seem to learn from the jumpers. why -- do not know. but it describes why software is what it is.
diana_coman: asciilifeform: neah, if they don't learn it's either because "not gory enough" or because "can't learn" (in which case fine, they are next to jump)
diana_coman: there is a lot of "not gory enough" around precisely because supported otherwise (usually on the strength of old stuff that still works basically)
diana_coman: ftr I suffered from the same fascination as shrysr , back when I was 18-22; the uni courses on "AI" were a lot of cold water (this?? is THIS what you call machine *learning*???)
diana_coman: and before shrysr thinks this was sometime in mesosoic, it wasn't.
shrysr: diana_coman:I agree. Hype is certainly very high. I've been at this for more than a year... trying to build a foundation of concepts. It has not been easy. However, my identification of this field -
diana_coman: shrysr: you got cut at "of this field - "
shrysr: on a fundamental basis is that i want to use a computer to analyse things.. build tools, workflows and systems to enable that. Over the years, i've found that my initial interest in compuational physics (fluid flow) was too niche and there were very few companies doing this. It was and is still too expensive for most.
shrysr: Especially after 'discovering' emacs, i've become more and more interested in code.
shrysr: irrespective of the hype, data science seemed that perfect combination of using code to derive insights and building systems.
shrysr: Otherwise >> I am currently implementing an ERP (enterprise resource planning) system for the company i work for. It will replace a completely manual system.
diana_coman: oh boy the ERPs
diana_coman: what are you building it in/on top of?
diana_coman: shrysr: eulora is from at least one perspective basically a laboratory for hands-on "data science" learning/perfecting/anything; and yes, you totally get to design and make your own tools to collect any data you think you need to figure out stuff (even *what* you should figure out is up to you!)
diana_coman: so how about this: get yourself an account in eulora, have a look around and choose something you want to model - from prices to quality or even quantity of crafting outputs
shrysr: I am using erpnext. It is actually very cool because i think it is the cheapest system out there, and one that trumps most mainstream offerings by the the fact that it is open source.
diana_coman: I guess I'll even offer a bounty for working model
diana_coman: if it comes to that (as so far it seems rather elusive :)) )
diana_coman: shrysr: on one hand I'm sure it's not at all hard to trump "most mainstream offerings" on anything really
shrysr: :D anything ??
shrysr: I'm checking out eulora right now
diana_coman: but for my curiosity: how many lines of code does this erpnext have and in what language? would you say you actually know exactly what each bit does and how and within what constraints/limitations?
diana_coman: shrysr: discuss it, you know? lolz
diana_coman: mainstream offerings are quite shitty; it quite comes with "mainstream"
shrysr: No. I haven't even explored the source code of erpnext, which is based on python. The focus has been on getting something working first. As such - it is maintained by a company, and we;ve signed on for support.
diana_coman: otherwise put you are slaving on a galley that is further enslaved by a bigger galley but the advantage of the whole thing is that ...it's open source!
shrysr: .. and cheap, easily 10x. I figured that if I reach a point where the software does not do what I want, i can atleast try to dig into the code and develop what I need.
diana_coman: shrysr: the trouble is that the moment you dig into it (and you'll get to that moment sooner rather than later, if you are doing anything serious as in actually useful there at all), you'll find that the code in question is such a collection of stacked chairs that developing what you need is not at all such a straightforward thing
diana_coman: eulora's client code is open source; and yes, pretty much was chosen precisely because "at least you can go in and do what you need"; it's been....a while, let's say and I'm still torching the swamp to then finally be able to do what I need
shrysr: i am inclined to agree. Besides the fact, that I have only started pushing myself to study the source code, the consultant I work with - often dissuades me immediately, when i talk of any 'deeper' customisation.
shrysr: I meant study source code in general.. not just for erpnext.
diana_coman: shrysr: a bit like with any other written word (literature would be the cannonical example), NOT all that is written is worth reading, that's the thing
diana_coman: and time is the only truly finite resource so don't waste it on reading shitty code unless you go in to burn it too at the very least
diana_coman: shrysr: re your consultant, no surprise really, think of it: he gets money out of you NOT knowing the stuff + him knowing just what he knows, not more; the "slaving on a galley" goes much deeper than a mere figure of speech
shrysr: I swear: on some days - my heart breaks with a fear that I will never understand and write elegant code. There is so much to learn... its infinitely worse when I forget what I've learned.
diana_coman: shrysr: that's the fear of any thinking man I'd say - that he'll never get to know what there is to know; on the bright side though, you have the advantage that there isn't all that much new elegant code written
diana_coman: anyways, eulora specifically *encourages* players to make/change their clients precisely as they want it - make bots to your heart content or anything else really
diana_coman: because yes, it's not a game for slaves on a galley but for thinking people
diana_coman: sure, the legacy code is a swamp but the advantage you have is that the earlier you get in, the better off you are over everybody that comes even later;
diana_coman: shrysr: re learning and if it's any consolation, I can add that in those past 3 years working on eulora I learnt way more (and deeper) than I did during my 3 years of PhD.
shrysr: ::D that is a consolation. 3 years sounds pretty quickkk. WHen i finished my masters - I just wanted to get out and work. I turned down a PhD offer in combustion .... once I started working, i realised I WANTED to do something original...I never did get another offer. I had deep regrets about that, but in the last few years - as I've 'discovered' github and the universe of 'code'... I've found peace in the fact that there is plenty to
shrysr: learn and do...that can help others, including original contributions.
diana_coman: hopefully you noticed also that github thrives precisely on fleecing the naive-but-well-intended who provide it with content for free and get pretty much jack-shit in return
diana_coman: don't fall for that; for one thing the best way you can help is anyway first of all by helping yourself too
diana_coman: and for the other thing, there's no reason to give your content to github; make yourself your own blog like a sane person and put it there; github is more than welcome to link to *you* if it wants to
diana_coman: (for that matter, same goes for dev.to , even in spades)
diana_coman: and yes, rather than run the risk of misleading youngsters to think it's fine and good to provide dev.to with their content, I'll refrain from it and I'll do the needed dig in their source rather
diana_coman: shrysr: Shreyas, eh? are you of Lithuanian descent by any chance?
shrysr: no!Well atleast as far as I know.. I am from India.
shrysr: Migrated to Canada in 2017
diana_coman: interesting; all Shreyas I knew so far are Lithuanians, ha; then again, Lithuanian has supposedly *something* in common with Sanscrit so...
shrysr: I have had plans to move my website to my VPS... but I guess I just settled for first succeeding in making content on a regular basis.
diana_coman: well, for all your web hosting needs, I warmly recommend Pizzaro: pizarroisp.net
shrysr: yes, my name has it's roots in sanskrit. Embarassingly enough, it's supposed to mean 'uniquely the best'
diana_coman: shrysr: so you have your work cut out for you there, by your very name :D
shrysr: my mom loses no opportunity to remind me
shrysr: :))
diana_coman likes the sound of shrysr's mum very much
diana_coman: shrysr: what's your website so far anyway?
diana_coman: honestly, why give github all that content; what if tomorrow they go down (like sourceforge did, remember?)
diana_coman: anyways, I'll be back later.
shrysr: Okay! Hey - wanted to say - it is so cool to connect with you.
diana_coman: hey, glad to hear that shrysr !
diana_coman: and possibly best set up znc or something and leave it on - nobody expects you to answer immediately, but at least you can see stuff later
shrysr: thanks for the tip on znc. Currently, I have emacs connected to erc on my vps and have logging enabled, so I should be able to login via ssh anywhere.
shrysr: Internet at my office sucks, so I was planning to look into mosh, based on advice in the emacs channel.
shrysr: Have to look into znc too.
shrysr: diana_coman:I was curious about the tools you use now? you mentioned dropping Emacs.
shrysr: diana_coman: re: content on github, I stumbled across https://sourcehut.org a while back, and never got fully set up. It does seem atleast a partial answer though?
← 2019-07-16 | 2019-07-18 →