The web in a box - a next generation web framework for the Perl programming language

IRC log for #mojo, 2016-05-16

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:50 sri_ joined #mojo
01:45 fildon_ joined #mojo
02:04 noganex_ joined #mojo
02:25 lluad joined #mojo
03:04 odc joined #mojo
04:34 sri the issue actually makes a really good case for job dependencies https://github.com/kraih/minion/issues/27
04:34 sri resizing 10 images in 10 parallel jobs followed by one upload job can't be such an uncommon case
05:48 inokenty-w joined #mojo
06:15 tyldis_ joined #mojo
06:20 tyldis joined #mojo
06:42 dod joined #mojo
06:47 dod joined #mojo
07:01 AndrewIsh joined #mojo
07:30 trone joined #mojo
08:23 icjs joined #mojo
08:28 Vandal joined #mojo
09:08 odc or you could just write one job with 10 forks inside it :p
09:11 meshl joined #mojo
10:19 andy joined #mojo
10:48 ncham joined #mojo
11:04 tchaves joined #mojo
11:08 dvinciguerra joined #mojo
11:15 tchaves joined #mojo
11:58 anon3252523 joined #mojo
12:07 nic joined #mojo
12:07 vytas joined #mojo
12:08 nic Catching up on the minion dependencies discussion
12:09 nic fwiw, the proposed language of 'parent'+'child' is very bad
12:11 nic As I understand it, jberger was proposing 'child' depends on 'parent', but that diametrically conflicts with two important uses of those terms
12:13 jberger nic: parent child in the graph sense not the process sense
12:14 nic We (computer science) often model job parts as dags or trees, so children are contributory parts of the parent
12:14 jberger Do you have some suggested terminology?
12:14 nic In project management and apps like RT, the completion of the parent depends on the completion of its children
12:15 nic In graph theory, the whole point of the parent-child language is that each child can point to unique parentage (many children, one parent)
12:16 nic But a job can depend on many other jobs
12:19 jberger I call it "parents" ;p
12:19 nic please don't; you will confuse anyone who tries to use it
12:20 jberger Then give me a better name
12:21 nic My own implementation uses 'depends' (I like brevity) but I saw depends_on was suggested earlier
12:21 jberger Then what do you call the jobs that are dependent on this one?
12:23 jberger Do you like ancestors/descendants better?
12:23 nic I don't, but the word 'dependents' would capture it fine
12:23 nic [answer to prev Q]
12:27 jberger My problem with depends(_on) dependencies dependents is that the directionality isn't very clear
12:28 jberger There is much more cognitive load in trying to understand those
12:28 nic The terms you were proposing have the opposite semantics to what our cognition expects
12:29 nic To me x depends y is clear, but others will prefer x depends_on y
12:30 nic If you need the reverse of depends_on, dependent_of is very clear
12:32 jberger depends/blocks i like better
12:33 nic yeah, that works for me
12:45 dod joined #mojo
12:47 CW joined #mojo
13:08 punter joined #mojo
13:15 elrey joined #mojo
13:17 elrey Hey guys. Does anyone know how can I create base class for my controllers, so I can reuse same functions for everyone?
13:19 batman elrey: https://metacpan.org/pod/Mojolicious#controller_class
13:25 bwf joined #mojo
13:29 jberger You don't even need that
13:30 jberger Just make a base class which inherits from Mojolicious::Controller and then inherit from that one in your controllers
13:32 gizmomathboy joined #mojo
13:33 jberger Though that is needed too i guess (for when you call build_controller or call a helper on the app)
13:33 elrey batman: is it going to load the specified controller with mojo's base controller? or replace it?
13:34 batman indeed. i think i misunderstood the question
13:34 batman elrey: no, for the existing controllers you need to do use Mojo::Base "MyApp::BaseController";
13:34 batman and then package MyApp::BaseController; use Mojo::Base "Mojolicious::Controller";
13:35 user_8822 joined #mojo
13:35 kaare joined #mojo
13:35 elrey batman: ok, ill try. thank you
13:36 user_8822 Hi. Is there a way to sort $self->render(json => { 'success' => $test2, 'errors' => \@errors });?  success is rendered below errors [ .. ] but I wish to have it at first for readability.
13:37 nic I quite like the syntax 'use MyApp::BaseController -base', but the official line (above) is more versatile/generic
13:38 nic user_8822: rename s/errors/wacky_errors/ and your problem goes away :D
13:39 jberger user_8822: hashes are unordered
13:40 jberger That said recently mojo json started emitting them sorted i thought
13:40 user_8822 I think it's sorted. Is there a way to distable it?
13:41 jberger If it weren't sorted it would be unordered
13:41 jberger So it would only be what you wanted some of the time
13:41 ilira joined #mojo
13:42 ilira joined #mojo
13:42 jberger And yes canonical json started in 6.58
13:43 user_8822 I don't think it's possible then.
13:45 sri jberger: what is blocks? that makes no sense to me
13:46 jberger A list of the jobs that it blocks from occurring, what we are this far calling children but nic dislikes
13:47 sri so the names parents/children are blockers for the feature now -.-
13:48 sri nic: why is a dependency tree with parents and children wrong?
13:48 jberger I can get behind depends(_on) as long as we can think of a word for the opposite
13:50 sri i guess previous and next would be more neutral
13:51 jberger Not my favorite but they could work
13:51 jberger Well
13:52 jberger Next isn't really because to me that says that will happen once this is done
13:52 jberger But a job may be waiting on several previous ones
13:53 jberger Next or Then would imply single parentage to me
13:54 jberger But it's just a name and that could be smoothed over with documentation
13:55 sri how complete is the patch?
13:56 jberger Honestly i got frustrated yesterday when i couldn't get a consistent benchmark
13:56 sri oh
13:56 jberger I still need to do the job_info query
13:57 gtodd I mostly use Mojolicious and sometimes Mojo::Pg  to connect to existing databases and write queries to grab info I want ... I haven't really used Mojolicious to create a full DB backed application the way Rails does for example (where you write ruby code that auto-creates a database for the app). I know Mojolicious isn't welded/bolted to database records in the same way but are there exampleson CPAN / github of "full stack" applications using a
13:57 sri i could do that too, it's not very important
13:57 gtodd I looked at applications like Galileo for inspiration of course :)
13:58 sri gtodd: https://github.com/kraih/mojo-pg/tree/master/examples/blog
13:58 jberger https://github.com/jberger/minion/tree/parents?files=1
13:58 jberger gtodd: Galileo is getting a little long in the tooth
13:58 icjs2 joined #mojo
13:58 jberger I'd do a lot different now
13:59 gtodd sri: thanks :-)
13:59 jberger Some of it is still good of course
13:59 gtodd jberger: heh already?  you should change it to use Mojoj::Pg ;-)
13:59 sri jberger: superjobs/subjobs ;p
14:00 jberger gtodd: it had a specific set of goals which it met but some of which weren't as practical as they were interesting
14:00 jberger It is trying very hard to be db neutral
14:00 jberger It is trying very hard to be cpan deployable
14:01 jberger It uses websockets for more than it should
14:02 gtodd right
14:02 jberger (Though developing Galileo did lead to several of the json over websockets infrastructure improvements that mojo got around then)
14:06 sri btw.
14:07 sri my knowledge of graph theory is not great, but even to me this sounds like bullshit ;p http://irclog.perlgeek.de/mojo/2016-05-16#i_12485469
14:07 sri what you're thinking of is a tree, not a graph
14:08 gtodd db neutral is good ... still I like the fact there is Mojo::Pg (de facto standard heehee) ,  Mojo::mysql ... since there might be a set of normal/popular/usually used patterns for doing things ...  of course something like  Rails is neat for bundling everything and all that but if you use Mojo you do get to have perl :) so you can connect to databases in lots of different ways
14:09 nic sri: you're being disingenous; a tree is a graph
14:09 nic s/ous/uous/
14:10 sri my point is that not every graph is a tree
14:10 nic we know
14:11 sri you're just wrong
14:15 gtodd jberger: just wondered if by looking at existing Mojo apps I could get a sense of what the standard/popular/usually used approach  might be. I was Googling around for "patterns" used in application development for web apps and databases I could compare (e.g. the use of "Active Records" is just one pattern or convention right?)  ... mojo-pg/tree/master/examples/blog will do for now :)
14:15 mcsnolte joined #mojo
14:16 gtodd jberger:  I did come across  https://metacpan.org/pod/ActiveRecord::Simple which was sort of surprising ...
14:20 genio My preference is a SQL::Abstract type of thing rather than ActiveRecord.  I might be in the minority here
14:20 jberger nic / sri: i believe you are both arguing in good faith
14:21 jberger We are all trying to make minion awesome
14:21 sri oh, i didn't realize nic was rude
14:21 sri so dingenuous means dishonest?
14:22 sri *disingenuous
14:23 jberger I'm just preemptively stating something i believe to be true
14:23 sri the graph databases i know don't even have the concept of directionality
14:24 sri there's just nodes with properties, and properties can reference any other node, how often they like
14:25 jberger These are at least dag-like right?
14:25 jberger Are there specific terms used in dag?
14:25 jberger I have to commute now but it is the shorter one today
14:25 jberger Biab
14:26 icjs joined #mojo
14:33 sri jberger: no, cycles are usually fine
14:45 sri interesting, the sidekiq guy has given up on his js port
14:49 jabberwok graph theory is surely one of the most under-appreciated, and under-used, branches of mathematics
14:49 jberger wow, that was fast
14:56 gtodd genio:  I have used that ... at the I thought of it as the "opposite" of then Active Record "pattern" but would never have been able to explain why :-)
14:56 itaipu joined #mojo
14:56 gtodd genio: It barely requires any other modules IIRC ... (Moo and .... ?)
15:00 genio just a few, yea
15:00 gtodd sri: jberger: why would sidekiq do that?
15:01 jabberwok agree with genio - the ActiveRecord way is "I don't want to design an actual database" whereas SQL::Abstract is "I want to get the actual database right, or someone has already done that; and I don't want to go nuts writing queries"
15:01 sri he didn't like the node.js culture it seems
15:01 orev joined #mojo
15:02 jberger I think I like not being in the culture that follows the "new hotness" language
15:02 gtodd jberger: :-D
15:03 genio More and more node.js is creeping it's way into our stuff here at $work
15:03 gtodd jberger: :-)  "perl ...  eww ... it looks like when you open a binary file in Notepad"
15:05 gtodd genio: you need hire people older than 22  :-)
15:05 preaction it's not an age thing, at all
15:05 icjs hi all, i'm using perlbrew in a Dockerfile but when i install a perl it backgrounds it - anyone know off top of head how i can stop it going into the background?
15:07 gtodd preaction: I meant more the dislike of perl ... but you are right ... it's more often folks who are set in the their ways that don't like perl and being set in one's ways correlates with ... well :)
15:08 preaction icjs: you might get better help in a more general perl channel. but, perlbrew blocks on the install, it's not backgrounded. do you just want to see the output as it's generated?
15:09 preaction gtodd: oh, i thought we were talking about the "new hotness"-following types. i think the arguments about perl's status in the larger software dev... community... are overdone and pointless :p
15:09 icjs preaction, ok i can try elsewhere. if i list the output as its generated that might be enough to keep Docker from not waiting for it to finish
15:10 preaction icjs: so docker just sees no output and assumes it's done? that sounds stupdi
15:10 gtodd anyway js can not be the new hotness for real ... it's 20 years olde  :)
15:11 jberger gtodd: server-side it is much younger than that
15:12 preaction accessibility is key. is why php won
15:13 icjs preaction, i think it's more that there needs to be something in the foreground always - i forget the exact technical reason for it.
15:13 gtodd jberger: :)
15:13 preaction icjs: something is. perlbrew is watching/waiting for the build to complete so that it can tell you whether it failed or not
15:14 icjs so if i can get it to block "fully" on the perl install then that might be enough
15:14 gtodd jberger: and we can haz perlito++ :)
15:14 jberger kinda
15:14 jberger perlito is cool and all but you can't really use it for real code
15:14 gtodd hmm but flavio's domain has been taken over ??!
15:15 preaction icjs: it _is_ blocked.
15:16 preaction icjs: https://gist.github.com/preaction/a3a7b523f00e1e6142676096268297e0 <- my perlbrew is now blocked. my prompt has not returned. the build will finish before i am allowed to run something in this terminal screen again
15:18 icjs preaction, hrm, thanks for that. let me check whats happeneing again...
15:19 zivester_ joined #mojo
15:22 dod joined #mojo
15:22 jberger icjs: perlbrew -v shows the output in the foreground
15:22 jberger IIRC
15:37 lluad joined #mojo
15:54 meshl joined #mojo
15:59 Vandal joined #mojo
16:02 PryMar56 joined #mojo
16:08 sri jberger: done some tests too
16:08 jberger do I have access on minion? I can't remember
16:08 sri for master i get about 4700 dequeues/s and with job dependencies it goes down to 4500 dequeues/s
16:09 jberger if so I can push my code as a branch to kraih/minion
16:09 jberger oh I though you meant unit tests
16:09 sri no, i meant benchmarks
16:09 jberger sri: interesting, that confirms some of my suspicions yesterday afternoon
16:10 jberger it was odd because I couldn't make the first commit cost-free either, which I thought we had done previously
16:10 jberger which is when I started tuning the benchmark to the point that it was more representative of the state of my laptop than anything else
16:10 jberger :-P
16:11 sri jberger: you didn't have access, but i've added you to minion and mojo-pg now
16:12 ramortegui joined #mojo
16:12 sri my benchmarks were fluctuating a bit too though
16:12 jberger cool, I'll make push what I have as a branch there
16:13 jberger https://github.com/kraih/minion/tree/parents
16:14 jberger that just makes it easier for you to hack on, if you want
16:14 sri jberger: looks like i can make it cost almost nothing though
16:14 jberger oh?
16:14 sri with a simple "parents = '{}'"
16:14 sri ;p
16:15 sri i assume the query planner knows how to optimize that better
16:15 jberger where?
16:15 jberger oh
16:15 jberger where parents = '{}'
16:16 sri "and (parents = '{}' or cardinality(parents) = ("
16:16 jberger nice
16:20 jberger I do see that in the list of alternatives now (as I google around) but interestingly most results from the postgresql mailing list seem to advocate array_upper or array_lower etc
16:20 jberger but go with what performs best!
16:21 jberger I tend to agree with you, that does look like something that the optimizer can understand
16:21 jberger complex function calls could be obscuring the meaning
16:32 PopeFelix Is there a way to define a single exception handler for all routes in a controller?
16:33 sri it's still about 100 dequeues/s slower
16:33 sri 4700 vs 4600
16:33 sri we'll have to decide how much job dependencies are worth for us
16:35 jberger 2%
16:35 jberger which is on a best case of zero work jobs
16:36 jberger given the number of requests we've had for this feature I'd say +1
16:36 jberger then again, I might be biased
16:38 jberger maybe time to ask the mailing list?
16:38 sri that didn't work in the past
16:39 preaction i mean, the reason we did this is because it was hard to build it with the existing tools Minion provides. so i'd say it's worth it. indeed, having this would make the job-queue thing i've had in my head for a few years basically trivial to accomplish with Minion
16:39 sri like i said, minion is in this weird state where there's not enough users to get good feedback, but too many to make crazy experiments :S
16:39 sri preaction: that is a good point
16:40 orev joined #mojo
16:40 sri stuff like expiring jobs is easy to tack on, but dependencies is really hard
16:44 sri crazy how wrong i was about the impact of skip locked
16:44 sri just removed it for lols, and benchmarks went from 4500 dequeues/s down to 2500
16:45 sri that's way past the 30% mark pretty much everyone expected
16:47 jberger :o
16:53 gryphon joined #mojo
16:53 trone joined #mojo
16:54 meshl joined #mojo
16:59 sri jberger: one more naming option predecessors/successors
16:59 * jberger chooses red paint for the shed
17:26 bwf LSF job scheduling uses the parent/child terminology : https://www.ibm.com/support/knowledgecenter/SSETD4_9.1.3/lsf_admin/job_dependency.dita
17:46 dod joined #mojo
17:51 andy joined #mojo
18:00 sri bwf: interesting
18:01 sri omg, there's now a game of thrones version of clue, i need this! http://gizmodo.com/everyone-is-probably-the-killer-in-game-of-thrones-clue-1776881522
18:02 sri HODOR!
18:03 andy joined #mojo
18:16 itaipu joined #mojo
18:17 jberger bwf++
18:18 jberger HODOR
18:24 gtodd trees have branches and leaves ... many also have fruit and nuts
18:26 vicash gtodd: they also have roots
18:26 gtodd true
18:27 gtodd some of course are conifers and have cones ...
18:31 gtodd so many network sysadmin script hacker job titles have "engineer" in them ...  that should make everyone feel safe and confident
18:39 jberger plicease: regarding your PRs to Mojo::Phantom
18:39 jberger thanks! I'll try to evaluate them tonight after $work
18:39 plicease sure np.
18:40 jberger I'm thinking that 4 and 5 can probably go in as is, and I'm thinking that maybe 6 should have conditional option?
18:40 jberger I think 5 makes sense since if the interpreter logs something I probably want to know about in the TAP stream
18:40 jberger I'm less certain about wanting to see all the page's log output
18:41 jberger it is set as note so it won't matter until you run prove -v
18:41 jberger but I wonder if there are some noisy pages out there
18:42 jberger don't have a strong opinion though
18:42 plicease yeah I did all of that stuff as note because i find it useful and not too noisy if I am using prove -v and expect more detail
18:42 plicease I am happy to make anything optional or not if that is your preference at $work there will probably be some kind of wrapper around phantom_ok
18:43 plicease so just being able to tweak through config should be more than sufficient.
18:43 jberger I can't decide :P
18:43 jberger you might be right, getting the firehose for -v may be best
18:43 plicease we could also make it default on, and have an option for off.
18:44 plicease that way you don't have my situation where you are frustrated for hours because you missed a simple diagnostic
18:44 plicease but you can turn it off if/when it becomes annoying.
18:44 jberger sure
18:46 jberger maybe these will help me diagnose the problem with #3 :-P
18:47 plicease one can always hope :)
18:53 nicomen I tend to use: some_test_ok or note diag $some_giant_error;
19:12 jabberwok vicash: do boxwood trees have square roots ?
19:27 vicash jabberwok: you're assuming a box is a cube for the roots to be square
19:28 * sri tends towards adding job dependencies to minion
19:29 sri all the new possibilities seem worth a 2% performance cost
19:30 sri but i assume the performance hit for ->repair will be more significant
19:31 vicash sri: i would think the performance cost is really subjective for the user. if the user is, like me, scheduling tasks that take more than 10-15 min to complete the 2% cost is not a concern, but if they're doing tasks that take milliseconds it may be of concern... either way I am looking forward to whatever solution you come up with
19:32 sri if a task takes milliseconds it doesn't belong in a job queue
19:34 marcus WINTERFELL!
19:34 sri SPOILERS ARE COMING
19:34 * marcus takes back the north
19:34 * genio dies of exposure
19:35 sri http://johnsonking.typepad.co.uk/.a/6a00d8341e442c53ef01bb084807ce970d-pi
19:35 marcus I'm not the prince that was promised, but I'm the prince you deserve
19:39 * jberger stabs marcus "for the watch"
19:40 sri FOR THE WATCH
19:42 marcus good thing it's pentecost.
19:44 Grinnz_ all this, over the watch? ... over watch? ... *twitch*
19:49 marcus iwatch?
19:49 sri more watch bands?
19:54 jberger FOR THE ROSE GOLD!
20:09 mishanti1 Was there ever a way to get mojo 3.x to not resolve symlinks when started with hypnotoad? I am trying to get zero downlime restarts working by having a symlinked path as the path I start hypnotoad with, and switching what the link points to when rolling out a new version.
20:09 mishanti1 But the apps process seem to be running with the "resolved" filename, and so the restart does not make the app use the new code.
20:11 mishanti1 Example: app is deployed to /path/to/app/app-<VERSION> . Symlink /path/to/app/CURRENT points to /path/to/app/app-<VERSION>. Hypnotoad is started with /path/to/app/CURRENT but ends up pointing to the versioned path.
20:12 punter joined #mojo
20:12 mishanti1 s/pointing to/running with processes using/
20:20 sri lol 3.x
20:20 sri i barely remember last week
20:22 mishanti1 Heh. Yeah, I know. It's old. We're slowly but steadily converting our codebase so the job of switching to 6.x is less of a pain.
20:22 marcus mishanti1: etoomuchtechnicaldebt
20:23 jberger mishanti1: admittedly 3.x is about when I joined, but I don't remember special consideration ever being given to symlinks
20:24 mishanti1 jberger: I had a peek at the code now, and it does indeed seem like what I want is not possible, as the paths get resolved during startup.
20:25 mishanti1 marcus: Admittedly some debt here in our codebase yes. But at least we have it under test. :)
20:25 vicash mishanti1: create a script instead of symlink that checks the config for the version and loads the appropriate version
20:28 mishanti1 vicash: I am not sure I understand what you mean. The script would find whatever pid the processes are running with, kill them and start hypnotoad again? If that is what you mean then yes, that would be my backup-approach if I can't get zero-downtime restarts working with the symlinks.
20:29 mishanti1 Huh. Ok, so it seems I can do it with Starman instead.
20:32 mishanti1 Or not. Tests failed. Oh well. Perhaps I can just fiddle with $0.
20:48 sri the minion feature list is pretty big
20:48 sri think the next time i have a few sponsored hours i'll write a minion guide explaining them
21:51 sri jberger: my first try https://gist.github.com/anonymous/0f8924c1d63666d101a64302e797fd35
21:52 sri just the docs are missing basically
21:52 sri even added jobs with unresolved deps to delayed_jobs in stats
21:56 jberger do you really need the cast on 'Parent went away'::text ?
21:57 jberger if so, man that is picky
21:58 jberger also, I take it array_length(..., 1) was more performant than cardinality? I guess that makes sense
21:58 sri yes, and cardinality is not really the right tool for the job
21:59 sri it counts nested arrays too, which feels a bit icky
21:59 sri but maybe that's just me
22:00 sri regarding repair performance, it costs 25%
22:00 sri 400 repairs/s down to 300
22:01 jberger even with no jobs with parents?
22:01 sri yes, it's still a table scan
22:01 jberger true
22:02 sri thought about merging it somehow with the worker went away check, but that's not good
22:04 sri haha, and now it's less than 25%
22:04 sri closer to 15%
22:04 sri ran a few dozen more times
22:08 jberger regarding array_length it just seemed odd to me to have to pick a dimension on a 1D array, but yeah, I'm ok with it
22:08 jberger anyway, I like the patch a lot
22:08 jberger +1
22:09 sri so, if i keep the ->finish in the benchmark, the overhead is barely noticeable
22:09 sri https://gist.github.com/anonymous/733b2e85b6dc9e18dbf60220eb175f33
22:10 sri just for repair
22:10 sri and a little for stats
22:10 jberger it'll be worth some publicity once (if) it gets released, I could write a perltricks article
22:11 punter joined #mojo
22:13 jberger as far as other backends go, SQLite should be able to emulate arrays by using json
22:13 jberger and maybe we should recommend that mysql backend die if parents is passed to enqueue
22:14 jberger http://dev.mysql.com/doc/refman/5.7/en/set.html ?
22:15 jberger that looks fragile
22:15 jberger Grinnz / bpmedley ^^
22:16 preaction it's also not the same. it's like ENUM, but allows multiple
22:16 Grinnz_ yeah, that's an actual set, not an array
22:16 preaction just a straight-up JSON type perhaps?
22:17 bpmedley jberger: i do not have mysql readily available..
22:17 Grinnz_ the only way to do arrays in mysql is to add a table :P
22:18 sri what the hell
22:18 sri now my benchmark reaches 2781 jobs per second
22:18 sri more than any version
22:18 sri hard to benchmark
22:19 mishanti1 sri: You're on a newer mac right?
22:19 sri would be funny if it was because the room temprerature decreased and the cpu overclocks more
22:20 Grinnz_ it is somewhat difficult to predict how new intel cpus will upclock/downclock
22:20 mishanti1 sri: There are issues with predictability of throttling on newer macs. More so than on the older models.
22:20 Grinnz_ you can keep it from switching somehow i think
22:20 sri mishanti1: i'm aware, even using the intel power gadget
22:21 mishanti1 Grinnz_: Not completely. You can control it to a point, but in the end the OS does have it's way.
22:22 sri best i can do is watch the cpu temperature i think
22:22 sri and make sure it matches for all runs
22:22 jberger I wonder if someone who has a vested interest in the mysql backend would like to help him (with bpmedley's consent)?
22:24 sri btw. don't think i can really measure a difference between array_length and cardinality
22:24 jberger use array_length
22:24 jberger it is the correct meaning
22:25 disputin joined #mojo
22:25 sri it's a line longer ;p
22:25 jberger hahahaha
22:25 * jberger doesn't care
22:28 sri mishanti1: the cpu is a core m7 btw. so an extreme case when it comes to throttling ;p
22:28 mishanti1 sri: Haha. Then yeah, extreme case for throttling.
22:29 sri doesn't get warm enough to throttle though
22:29 sri even under heavy load it hover at 50C
22:29 mishanti1 They random-throttle. It is one of their issues.
22:29 sri +s
22:29 sri that was the old ones
22:29 sri this is a skylake m7
22:29 mishanti1 Still in the new ones sadly. :(
22:29 sri nope
22:30 sri nothing random about it at all in my tests
22:32 sri even under heavy load and 90C it can keep the cpu at 2.5ghz steady
22:33 sri which i think is the max for multi-core workloads
22:35 sri mishanti1: got a link for your information?
22:35 * sri has actually been reading a little on reddit about the new macbooks and core m over the last few weeks... and there was soooo much bad information ;p
22:38 disputin joined #mojo
22:38 sri this feels like a 15W i7 cpu pretty much, if you wanted to look for weaknesses i'd say the gpu maybe, which seems more flaky
22:40 mishanti1 sri: Yeah, there will always be a lot of FUD floating around, and on reddit especially. I've been hearing about it on youtube channels of computer-repair shops.
22:41 mishanti1 sri: If you're curious, Louis Rossmann on youtube has some great technical information on the inner workings of (especially) Apple computers.
22:41 mishanti1 He's a repair-guy (sometimes a bit foul-mouthed when he's fired up about something) but has some very decent technical insights.
22:41 mishanti1 https://www.youtube.com/user/rossmanngroup
22:45 sri don't see anything about the new macbook
22:47 mishanti1 sri: One issue with his videos is that they are not very "on topic", so he might mention/talk about a lot of stuff but not put that in the description or title. I just put them on and listen to it while working, since they are crazy long.
22:48 mishanti1 sri: So no, I sadly do not have any links to back up my previous statement with. Just remember having heard people like him complain about it.
22:56 sri opened one video with macbook in the title, watched him repeat the same silly rant 50 times... never gonna watch that channel again ;p
22:57 sri the time of repairable computer is gone, he needs to get over it
22:57 sri +s
22:58 mishanti1 sri: Uhm. Those days are in fact not at all gone. There are often pretty simple things that fail, and that are easy to repair.
23:04 Adura I repair by replacing parts that are acting up.
23:04 Adura Though, doing that with a laptop would be a bit harder...
23:07 mishanti1 Yeah, need a bit of equipment to do that.
23:08 disputin joined #mojo
23:12 tchaves joined #mojo
23:25 andy joined #mojo
23:40 bpmedley jberger: I'd gratefully give consent for MySQL backend.  I just need to be told what to do.

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary