Recent Forum Posts
From categories:
page 1123...next »
hmf (guest) 11 Dec 2019 13:55
in discussion Hidden / Per page discussions » Unikernels: No Longer an Academic Exercise

Is a docker container not very much like a unikernel on a more coarse grained level?

by hmf (guest), 11 Dec 2019 13:55
Softgen Infotech (guest) 11 Dec 2019 11:55
in discussion Hidden / Per page discussions » Tiles: Report on Programmatic Code Generation

Looking for Best Training Institute in Bangalore , India. Softgen Infotech is the best one to offers 85+ computer training courses including IT Software Course in Bangalore , India. Also, it provides placement assistance service in Bangalore for IT. http://www.softgeninfotech.com/#

by Softgen Infotech (guest), 11 Dec 2019 11:55
Apostolis (guest) 07 Dec 2019 22:48
in discussion Hidden / Per page discussions » The Missing Piece

I personally use Agda as a specification language, which is pure. If we model organizations with Agda, and our model of the organization is precise, it will be similar (or identical) to a pure function.

This means that if the environment changes but the input remains the same, the organization will behave the same.

Specifying the exact causes / inputs that affect an organization can let us replicate the organization to different environments and have the same predictable outcome.

by Apostolis (guest), 07 Dec 2019 22:48
robert (guest) 07 Dec 2019 17:51
in discussion Hidden / Per page discussions » The Missing Piece

Can you elaborate on your comparison with referential transparency / side effects ? I don't see how they relate to the matter at hand.

by robert (guest), 07 Dec 2019 17:51
eTechno Soft Solutions (guest) 07 Dec 2019 11:46
in discussion Hidden / Per page discussions » Tiles: Report on Programmatic Code Generation

Looking for SAP FICO Training in Bangalore, learn from eTechno Soft Solutions SAP HR HCM Training on online training and classroom training. Join today! To learn more click here- http://www.etechnosoft.in/sap-fico-training-institute-in-bangalore.html

by eTechno Soft Solutions (guest), 07 Dec 2019 11:46
Apostolis (guest) 06 Dec 2019 19:59
in discussion Hidden / Per page discussions » The Missing Piece

We do not need to know what is inside people's minds. We have a specification for that, called a job description. If you are qualified, then you can perform similarly to anyone else who qualifies.

For the same reason, we do not care how the cpu has been manufactured or how it performs its arithmetic operations, as long as it abides by the specification.

The main problem is to have the right level of abstraction. If for example, in a cpu, we ignore endianness , then we will get different results for the same program if it is executed in different cpus.

Having the precise specification of an organization helps in avoiding that problem.
Functional programmers call this referential transparency, or working without side-effects.

by Apostolis (guest), 06 Dec 2019 19:59
Softgen Infotech (guest) 05 Dec 2019 16:00
in discussion Hidden / Per page discussions » Tiles: Report on Programmatic Code Generation

wonderful thanks for sharing an amazing idea. keep it…

Softgen Infotech have the best Python Training in Bangalore . Any professional who is looking out to switch their career can enroll with us. To know more click here- http://www.softgeninfotech.com/Python-Training-Institute-in-Bangalore.html

by Softgen Infotech (guest), 05 Dec 2019 16:00

No. Linus Torvalds "real argument" is not becuase he doesn't know Linux.

Linus works on a (large) kernel project. C is the BEST language for this job. However, the real reason why he does not use C++ is because he has many contributors… and many of them can submit poor C++ code… which is harder to reason compared to poor C code.

If you could be bothered to read he comment in full, you would understand that.

by msmith (guest), 04 Dec 2019 12:49
robert (guest) 03 Dec 2019 14:03
in discussion Hidden / Per page discussions » The Missing Piece

Apostolis you are looking at it from the wrong perspective. Humans are the most important factor, not organizations. "Forking and mutating organizations" would fail just as the CAR-Switzerland situation fails. You need to capture human nature, human mind, not the mere creations produced by it (such as organizations). Even if you managed to digitize "social organization DNA", you still wouldn't have the complete picture; you would miss the most important factor — the human as individual. And if you then took a 'good' organization which is efficient, well-defined, working nicely (eg. the Swiss economy) and gave it to the wrong individuals (eg. the CAR government), it just wouldn't work. And since you cannot replicate human nature, or force Central Africans to think and behave like the Swiss, your effort is, I'm afraid, vain.

by robert (guest), 03 Dec 2019 14:03
mentalisttraceur (guest) 23 Nov 2019 06:27
in discussion Hidden / Per page discussions » Select Statement Considered Harmful

One case I don't see obviously cleanly handled by this model, either here or in the follow-up footnote post, is that of collating multiple data sources into different updates to the same state.

Actually I just thought of a solution when I came back after a few hours to write this comment, so this does fit into the model, but I'll write the rest of the post for anyone else who it helps.

For example, I have a program right now which receives information from many different data sources about some networked devices we have. (The data sources are various systems that directly or indirectly talk to our networked devices, or have insights into the devices or their activity for other reasons.)

You can picture each data source as a Go or libdill -style channel, where events arrive frequently - nearly constantly - but effectively at random.

Then each of these pieces of information needs to be parsed (different for each source) into a decision about how to modify the in-memory state for each of those networked devices (the state is shared).

At first glance, this is not a "semantically the same" data sources problem, so it cannot be handled as one channel.

At first glance, this can be trivially split into independent coroutines, but it relies on sharing of state and - assuming that the concurrency can also potentially be parallel at any moment, as in Go - requires bringing mutexes into the equation. That seems strictly worse for readability because now the state manipulations are not exactly local in the code.

So I kept feeling like a select on all of these data source channels was the cleanest way, that kept the semantically related code local in the source text.

However, it finally hit me:

1. Decision on if and how to mutate shared state
2. Mutating shared state

These are two semantically separate things, so they actually fit this no-select-statement model beautifully.

Separate coroutines can read each data source channel, transform the data into some representation of the state change to make, and then write that representation into a single shared state updates channel.

A single coroutine reads just that channel and applies each state change.

Evidence through concilience that this is actually a good idea - doing this automatically factors the logic into more easily testable pieces - the state changes are either data which can be fed into the logic that applies them, or functions which are given the state to update as an argument - each of which lends itself way more easily to unit testing, fuzzing, model-based testing, etc.

You're onto something real and significant here Sustrik.

by mentalisttraceur (guest), 23 Nov 2019 06:27
mentalisttraceur (guest) 23 Nov 2019 05:47
in discussion Hidden / Per page discussions » Select Statement Considered Harmful

I think these two problems are easily addressed:

1. If the underlying system deeply supports timeouts or cancellations - at every blocking operation or context switch for example - then you *never* have to remember to add a timeout parameter - callers can just impose timeouts on every operation within some scope wherever appropriate. Both this blog author's libdill in C and the very nice trio library in Python are examples of this - assuming the entire system is written on top of primitives that can be timedout or cancelled, timeouts come for free, everywhere.

Which actually gets you the timeouts handled orthogonally that you wanted, just like Go's select on a timeout channel does.

2. This is actually solved by the same way that I described in the previous point - if the underlying primitives can be timed out or cancelled in a scoped way then you get that feature for free.

But if you didn't, that's actually a really trivial pattern to abstract - I have a library that boils down to like ten lines of essential code which lets you do something like (Go-like pseudocode):

timeout := totaltimeout.Timeout(5 * time.Seconds)
someOperation1(…, timeout.TimeLeft())
someOperation2(…, timeout.TimeLeft())

someOperationN(…, timeout.TimeLeft())

and so long as your operations handle zero for the timeout as a "bail immediately" (which can be trivially ensured with a wrapper function if they don't), it works exactly as intended.

by mentalisttraceur (guest), 23 Nov 2019 05:47
mentalisttraceur (guest) 23 Nov 2019 05:12
in discussion Hidden / Per page discussions » Select Statement Considered Harmful

To me that statement about Go being the closest to the UNIX philosophy was self-evidently true the moment I read it.

It's actually insightful in a deep way, but I think rmoe is reading a different meaning into it, a more value-judgement-feely meaning instead of the common-pattern-identifying meaning, probably because the design decision parallels and similarities are somewhat "gated truths" - it's only obvious to me what you meant because I have spent a really large amount of time thinking about the relevant design considerations, the underlying unifying base principles on top of which a lot of systems and languages were built.

I think rmoe is right in the sense that obviously those who don't already "see it" are liable to misinterpret that statement or have a negative reaction to it, just like rmoe did. But it's not an absolute, and if you're writing for the audience of people who already get it or are willing to do good-faith idea-fitting on the statement, taking the time to justify it has limited value, and from experience I can say even counter-productive.

So I'm here to say the statement was good, right, and while I'm sympathetic to rmoe's critique, I am glad you just wrote it simply at face value and moved on - it worked well for me.

by mentalisttraceur (guest), 23 Nov 2019 05:12
Robert (guest) 21 Nov 2019 22:23
in discussion Hidden / Per page discussions » The Politics of Age (the Young vs. the Old)

One of the primary arguments for 18 and not older was because that's when you can be drafted. Clearly that argument doesn't apply to women but they got the benefit as well. Also 18 is a legal adult, able to do pretty much anything without a co-signer/guardian so if the voting age was raised that would be inconsistent unless those were raised as well.

I'd sooner consider a maximum voting age (say 85 or life expectancy + n where n is probably ~5) than lowering the voting age below 18.

by Robert (guest), 21 Nov 2019 22:23
Apostolis (guest) 04 Nov 2019 14:06
in discussion Hidden / Per page discussions » The Missing Piece

In societies , there is no accurate replication mechanism. Most organizational information is either private property or implicit. For this reason, I am trying to digitize the social organization DNA. Another reason is efficiency. To easily fork and mutate organizations , in the same way we fork code.

by Apostolis (guest), 04 Nov 2019 14:06
watson (guest) 25 Oct 2019 06:31
in discussion Hidden / Per page discussions » Tiles: Report on Programmatic Code Generation

Thank you for providing this kind of useful information,I am searching for this kind of useful information; it is very useful to me and some other looking for it. It is very helpful to who are searching Selenium Training.
http://www.ecaretechnologies.info/Selenium-Training-Institutes-in-Bangalore.html

by watson (guest), 25 Oct 2019 06:31
watson (guest) 25 Oct 2019 06:30
in discussion Hidden / Per page discussions » Tiles: Report on Programmatic Code Generation

Wow it is really wonderful and awesome thus it is veWow it is really wonderful and awesome thus it is very much useful for me to understand many concepts and helped me a lot. it is really explainable very well and i got more information from your site.ry much useful for me to understand many concepts and helped me a lot. it is really explainable very well and i got more information from your site.<a href="http://www.ecaretechnologies.info/Selenium-Training-Institutes-in-Bangalore.html">Selenium training in bangalore</a>

by watson (guest), 25 Oct 2019 06:30
Apostolis (guest) 18 Oct 2019 20:20
in discussion Hidden / Per page discussions » On Becoming Clueless

There is plenty of work that does not require intellect. In fact , knowledge-based jobs are a small percentage (3%) of total jobs. (researchers and information workers)

1. http://www.bls.gov/web/empsit/ceshighlights.pdf
2. http://data.worldbank.org/indicator/SL.TLF.TOTL.IN?locations=US

by Apostolis (guest), 18 Oct 2019 20:20
Evgenii (guest) 02 Sep 2019 10:26
in discussion Hidden / Per page discussions » Anti-social Punishment

Hi all!

I want to add to a previous comment with a joke about farmers and cows.

There is a joking saying in Ukraine and Russia, that happiness is not when you bought a cow, but when your neigbour`s cow died. : )

by Evgenii (guest), 02 Sep 2019 10:26
Dorkus Malorkus (guest) 13 Aug 2019 02:32
in discussion Hidden / Per page discussions » Linguistics and Programming Languages

In Python (3.7), " if not foo == 'A' or 'X' " will return True no matter what the value of foo is (if any).

This is because the condition (foo == 'A'), as well as ('X') respectively, are evaluated separately (and in turn).

All it takes for an OR clause to pass in Python is a single True somewhere along the road during short-circuit evaluation (see the wikipedia article on 'short-circuit evaluation').

by Dorkus Malorkus (guest), 13 Aug 2019 02:32
Yuriy (guest) 11 Aug 2019 14:22
in discussion Hidden / Per page discussions » Hard Things in Computer Science: Naming things

This is a good example of context-specific semantics. You only understand what it is because it's so common within the context of OOP. If it weren't for GoF and the popularization of "Design Patterns" due to them, you would have pretty much no idea of what a "Factory" is. So we need a common framework in which we think about the things, and consequently name these things. OOP design patterns is an example of such a framework. Obviously such an approach has vast advantages but can be quite limiting, especially when not thought out carefully.

by Yuriy (guest), 11 Aug 2019 14:22
page 1123...next »
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License