#ossasepia Logs for 15 Aug 2019



April 19th, 2020 by Diana Coman
shrysr: Re (1) : I did not 'run away' knees knocking in fear of v.py to docker, and don't understand the reference to making noise, unless you are calling it worthless, which I would disagree with. Yes, docker does some hard stuff, but so that I can do harder stuff, and work backwards when possible. In any case - it doesn't do something I've not struggled with?! BTW, I was reflecting that I could probably take [00:16]
shrysr: apart a smallish motorbike (except the engine). Today, I was learning about make files, reproducible research and will have a nice container up soon, and intend to use to enhance the documentation of a package to begin with. Another (similar to UMAP) problem popped up today - different results on different machines with supposedly the same algorithm and data (in the course community). Took up docker [00:16]
shrysr: because of several reasons, a bunch of which were in fact shared in the channel and should be mentioned in the posts as well: to produce reproducible results, and to construct my own docker toolbox to facilitate the above, as well as experimentation with different algorithms using a consistent and selected version of R and packages.. quickly evaluate / experiment to try resolve the problems mentioned [00:16]
shrysr: above. Beyond this - i prefer to use docker to host shiny apps. Shiny is a R package to build web apps. Its a cool way to showcase projects, and also to build tools at work. The above is supposed to be parts of a plan, atleast. [00:16]
shrysr: Re (4) : The website 'futzing' falls under the 'random emotion dominated jump' category, and could have been postponed as I had a working setup - but the end result is fortunately quite pleasing (to me) nevertheless and there were lateral benefits. However, the 'switch' to Docker was not random as it was already existing, as a part of a bigger plan. I believe I did say this? In fact, docker and v.py was [00:17]
shrysr: kindda the plan. [00:17]
shrysr: While I've come here to find meaning and like what you have to say and think I have the stomach for 'the red pill truth' and want to work; there also happen to be some immediate considerations for me, like - learning skills / concepts / developing a project portfolio which will get me to the next 'better salt mine', which I needed several months ago. I don't yet know what TMSR is, and I do not know where [00:17]
shrysr: this is headed, and obviously anything worth doing will take effort and time, which I will do - because I want to follow you. Just so you know - I did no background search on you before rating you 10. I believe in you, as my teacher more out of instinct and curiosity at this point and a very superficial glance at your profile - but the fact is that I need to be able to confidently state that I've 'herped [00:17]
shrysr: and derped' Docker and MANy more XYZ. Besides this - i am still 'herp derped' on my 'salt-mine slave' shit. All this being said - I think I need to adhere to (5) - X lines of code, take it slower and smaller. [00:17]
shrysr: Re (4) : I understand what you mean about keeping deadlines and appreciate the value of *any* constructive advice or attention given to me - and have written that I should have communicated that I'm working on something else - most of which is written in the summary I see. However, considering my flimsy reply earlier to lobbes comment as well as the fact that nobody here knows me + the fact that I am still [00:19]
shrysr: quite ignorant about what TMSR does / is and proabably rant 'I'm a lost headless chicken' a little too much: this is not the the first undertaking of my life, and in the past - while I did not 'change the world', and the solutions were nowhere close to being paragons of coding elegance - I have been able to single handedly drive and implement systems and solutions that had a significant impact, ahead of [00:19]
shrysr: time, repeatedly, among other things. At the end of 5.5 years - I made +5X the money I started with, purely based on performance and impact, in 2 completely different organisations, after which I shifted here. I'm not stating the length of my dick in any sense (though it was overall a good achievement in several respects) - only trying to convey that I am not and have not been utterly devoid of [00:19]
shrysr: professional standards nor am I ignorant of the impact of sticking to a deadline. Also in continuation to my Re (4) above - i do not yet see fit to abandon the things I was working on, before popping up in this channel... because I'm not in the situation wherein "today's 'salt mine' is quite acceptable and I can spend all my time searching more meaning in life". I still dislike sharing such details with [00:19]
shrysr: people I do not know, especially in a public, perpetually logged forum, and I questioned the need to explain all this - Nevertheless - I just did, perhaps more out of a interest to convey that I am not a worthless escape artist / vagabound - and a little bit about wanting to share it with you. [00:19]
shrysr: Re (3) : Okay for reading the article + summarising without any take. Re: causes and purposes - what exactly was the idea I utterly missed? I referenced a specific paragraph of the article talking about theoretically making decisions only based on cause and not allowing purpose to influence - and wrote about my 'struggle' in dealing with a 'doctrine' esque cause. I tried to say it is not practically [00:19]
shrysr: possible to follow a cause without a cost that you may be unwilling to pay in the present, or that it is not possible to eliminate purpose from a decision by virtue of being human in a society today. There is no control over purpose and the future, agreed - but it creeps into our decisions and I questioned whether it need be viwed as an impediment. It *was* intended to be my take, and not a summary of the [00:19]
shrysr: reference. If that was not well written - ok - I want to, and I will certainly improve, but I don't see how this message was entirely misplaced in the article, and neither can see it being so utterly devoid of connection to the reference article. Re: WoT - i need more exposure to the WOT and actually use it to understand better. [00:19]
shrysr: Acknowledgement: the reasons are addressed above. As such, the website 'futzing', though categorised random was also done to make it easier to post notes and summarise weekly, which I conveyed, to me it is a measure, though you probably don't see the value in that (either?). In reality, I'd forgotten that I had already written a summary a few days before and actually could have published atleast one [00:20]
shrysr: summary earlier. Atleast one more measure is that I've fixed locations/workflows for the notes I publish. Re (5): ... all or nothing. Yes. I have strong tendencies to forget this... it is a problem. I was reflecting last night after the summary that it would be 'relieving' to reduce the magnitude and have shorter, consistent bursts. A target like X lines certainly lowers 'barriers' in my head. [00:20]
shrysr: Deadline - Sunday. For all of it. .. [00:20]
diana_alt: shrysr: nice outburst there :) [03:59]
diana_alt: beats keeping silent by a *large* margin, you know? [03:59]
diana_alt: that being said, it is way better to consistently keep in sync as it were rather than bursting when prodded more strongly [04:00]
diana_alt: shrysr: and you do seem to fight some strawmen there (e.g. I said you ran away from v.py because for a few weeks now *all* that comes out re v.py is that you find it hard; but nothing concrete, no "here's what I tried and where I got stuck", no "what does this do because I think it does X but Y", no nothing); onth during this same time I heard lots of all sorts that were not even mentioned as "my [04:08]
diana_alt: plan for this week" [04:09]
diana_alt: shrysr: do me a favour and give me a link to the topics/issues you are going to spend this week on; basically the currently missing point 4 in http://younghands.club/2019/08/14/review-summaries-week-3-and-4/ [04:10]
diana_alt: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000164 -> so put it explicitly and in some more detail in the plan, what! [04:13]
snsabot: Logged on 2019-08-15 00:17:47 shrysr: While I've come here to find meaning and like what you have to say and think I have the stomach for 'the red pill truth' and want to work; there also happen to be some immediate considerations for me, like - learning skills / concepts / developing a project portfolio which will get me to the next 'better salt mine', which I needed several months ago. I don't yet know what TMSR is, and I do not know where [04:13]
diana_alt: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000165 -> why exactly can't you confidently state? if you mean that I should approve /refrain from pointing out trouble *because* it lowers your confidence in stating/sends you into hiding, that's a. not helping you at all b. not going to happen. [04:16]
snsabot: Logged on 2019-08-15 00:17:49 shrysr: this is headed, and obviously anything worth doing will take effort and time, which I will do - because I want to follow you. Just so you know - I did no background search on you before rating you 10. I believe in you, as my teacher more out of instinct and curiosity at this point and a very superficial glance at your profile - but the fact is that I need to be able to confidently state that I've 'herped [04:16]
diana_alt: shrysr: understand that all the above and previous feedback is strictly what the words say, don't make strawmen out of it; in particular: if I point out where you failed, it does NOT follow that you didn't do anything RIGHT! and yes, I am quite sure that you *CAN* do things right and even very well otherwise - that's precisely WHY I keep you to a good standard and simply point out that so far re [04:23]
diana_alt: deadlines and feedback here you repeatedely failed (most probably out of too much enthusiasm if that makes you feel better about it, I am aware how easily that can happen but it is what it is.) [04:23]
diana_alt: re Docker: I said it makes noise because it brings with it a *lot* of dependencies and assorted things that a. will weigh you down b. you have no control whatsoever over c. sooner or later (and usually sooner) *will* break from under you and as a result force you to "upgrade" all sorts + import yet another spittoon of neverending shit. This is neither specifically Docker btw not news as such. [04:29]
diana_alt: the spittoon is a reference to "it's all one strand" illustrated with a great deal of matching-crudeness in http://www.loper-os.org/pub/the_spittoon.txt [04:30]
diana_alt: if you say that you *have to* drink this particular spittoon because of current/planned salt mine, it's your decision and stand by it for what it is. I'm simply making sure that you are aware - as much as possible - of what you are getting yourself into there. In other words, it flows from causes, not purposes: I'm saying it *because* I've seen it all before and I know the problems with it and [04:34]
diana_alt: *therefore* I can't keep quiet ("so you are confident in sharing" or whatever) and let you get sucked into it without even realising what's going on. [04:34]
diana_alt: I'm not saying it towards a purpose e.g. "so that he drops it" ; sure, I wish you were at the point where you decided to not waste time on it but if you are not there, you are not there and that's all. [04:36]
diana_alt: and the above being said, even dives in swamps can be useful learning (as long as you don't drown in there), sure; there is some potential trouble in that you may end up learning things to unlearn later too but such is life - things are neither ideal nor "as they should be" or something. [04:42]
shrysr: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000180 Yes. [12:11]
snsabot: Logged on 2019-08-15 04:00:33 diana_alt: that being said, it is way better to consistently keep in sync as it were rather than bursting when prodded more strongly [12:11]
shrysr: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000186 No I did not mean so and agree that would be counter-productive. [12:11]
snsabot: Logged on 2019-08-15 04:16:14 diana_alt: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000165 -> why exactly can't you confidently state? if you mean that I should approve /refrain from pointing out trouble *because* it lowers your confidence in stating/sends you into hiding, that's a. not helping you at all b. not going to happen. [12:11]
shrysr: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000191 That was a disgustingly fleshed out, but weirdly instructive story which made me laugh (a bit) and cringe (a lot). I'm not sure if I'll eat anything for awhile, but I get what you meant. That being said, - I can't think of any non-docker solution to a reproducible environment other than creating a 'frozen version', preferably under version [12:11]
snsabot: Logged on 2019-08-15 04:30:22 diana_alt: the spittoon is a reference to "it's all one strand" illustrated with a great deal of matching-crudeness in http://www.loper-os.org/pub/the_spittoon.txt [12:11]
shrysr: control. Can do that Immediately, lol. I used to do that with Emacs packages, but I'm not sure if it would work as intended with R across platforms, I think the paper below mentions somewhere it cant be. I thought the fundamental premise of using docker was to ensure/guarantee reproducibility of the computational env, in fact, just about to finish: https://peerj.com/preprints/3192/ , and atleast till pg [12:11]
shrysr: 19 - nothing is said about potential pitfalls. [12:12]
shrysr: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000188 Okay. I see what you mean. In general - I was reflecting that I seem to talk it out in reasonable detail my head, perhaps even all day, but not enough makes it to the screen / IRC / notes to justify the thread, assuming there was any validity to the thread to begin with. Or, too much makes it on screen bringing straw men to life. I've been thinkin [12:12]
snsabot: Logged on 2019-08-15 04:23:37 diana_alt: shrysr: understand that all the above and previous feedback is strictly what the words say, don't make strawmen out of it; in particular: if I point out where you failed, it does NOT follow that you didn't do anything RIGHT! and yes, I am quite sure that you *CAN* do things right and even very well otherwise - that's precisely WHY I keep you to a good standard and simply point out that so far re [12:12]
shrysr: I should schedule time-outs where I just take a deep breath, pause and unravel the threads to assess where I am w.r.t the starting point. [12:12]
shrysr: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000194 [12:13]
snsabot: Logged on 2019-08-15 04:36:15 diana_alt: I'm not saying it towards a purpose e.g. "so that he drops it" ; sure, I wish you were at the point where you decided to not waste time on it but if you are not there, you are not there and that's all. [12:13]
shrysr: - This is hard to answer. I think I know, and I've had guidance in general from people I do trust. But there's a lot of 'variety' in experience and too much superficial show and tell (that is applauded to the roof WTF), and "data scientist" means a different things in different places, and few admit their approach to *not* be 'universal' or piled on a stack of unstable chairs. I also see deficiencies in [12:13]
shrysr: the way people handle data science (both experienced and newbies) - they just 'use' algorithms and talk 'Business value, ROI' etc, and *often* avoid going in-depth in a lot of areas that I think are no less critical - it is a fucking science. I want to know ATLEAST the intuition of the math (preferably all of it) before using the algo... it has slowed me down very much and has needed more 'effort' and [12:13]
shrysr: energy to stay consistent / stable / motivated. I think that - if I drop all my efforts to learn and thus include these 'catchphrases' in my resume - i.e I take up v.py full time - it means staying in current salt mine for Atleast close to a year minimum before I 'find my feet' to the extent of landing another salt mine. It may well still take that long at my current rate of progress (if did not do v.py [12:13]
shrysr: full time). I think that staying here is dangerous because I work completely alone and its IMHO among the worst culture/admin/attitude i've seen... shitty pay etc. Not that I was expecting paradise - but there almost no intellectual camaraderie. I have Tonnes of freedom by virtue of working alone, but not probably not the kind conducive to implementing 'cool shit'. As such the ERP thing is not terrible... [12:13]
shrysr: but a lot of manipulation was done to reach that point, over months. [12:13]
shrysr: bblx01 [12:14]
asciilifeform: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000200 << shrysr , the v.py you've been studiously avoiding, is in fact that 'non-docker solution to reproducible environment' . [12:19]
snsabot: Logged on 2019-08-15 12:11:56 shrysr: http://logs.nosuchlabs.com/log/ossasepia/2019-08-15#1000191 That was a disgustingly fleshed out, but weirdly instructive story which made me laugh (a bit) and cringe (a lot). I'm not sure if I'll eat anything for awhile, but I get what you meant. That being said, - I can't think of any non-docker solution to a reproducible environment other than creating a 'frozen version', preferably under version [12:19]

Comments feed: RSS 2.0

4 Responses to “#ossasepia Logs for 15 Aug 2019”

  1. #ossasepia Logs for 15 Aug 2019
    April 19th, 2020

    I ended up here by clicking this link, should be Sept 7th rather ? Something got broken here or wth, I'm now confused.

  2. Diana Coman says:

    Hm, not sure what link you mean - the one to http://logs.ossasepia.com/log/trilema/2019-09-07#1934510 ? That would be to the old shifty-sands logger that I decommissioned indeed because wtf to do now. The chan there being #trilema though, I would have expected that link to point now to your logs-on-the-blog page, no?

    At any rate, for #ossasepia and #eulora logs, I published the whole archive of direct matches old/new link (since sadly there is no direct rule to fit all cases - I didn't think of fiddling the publishing date of the blog versions right from the start, unfortunately).

  3. Diana Coman says:

    Uhm, I gave the wrong link there, sorry, that's the archive of actual log contents. The matches/explanations are here: http://ossasepia.com/2020/06/16/updating-your-old-log-links/

  4. A damn, I see I missed some things here.

Leave a Reply to Diana Coman