I'm a novice mturk user. I created HITs for crowdsourcing using external question hosted on a server. I wanted to know if there is a web interface where I can see progress of my HITs. I tried looking at the https://requester.mturk.com/manage and https://requestersandbox.mturk.com/manage. But I cannot see the HITs programatically created using boto3. Should I look somewhere else? If not what's the way to get this information?
I share your pain right now. As of June 2020, this situation hasn't changed. HITs that are NOT created through the MTurk web interface STILL do not display on the web interface. It's terrible. We have 3 options for seeing and managing the HITs:
Use scripting and boto3. <-- Best option for now.
Use the AWS CLI.
Use the AWS shell (aws-shell).
I think the best option is to make scripts that do exactly what you need. Chances are you'll need to do things more efficiently than you could using the AWS CLI only. aws-shell isn't easy enough to use, and it also looks unsupported for over a year at this point (judging by their official github issue tracker).
For what you're asking specifically you'll need to use the method list_hits() and possibly list_assignments_for_hit(). See https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/mturk.html
Also I'm very new to this, so if it sounds like a barely or only sorta know what I'm talking about, that's correct. But I also wished there had been a straightforward answer to this question a couple weeks ago when I was sitting here dumbfounded.
Related
Is there a way to specify the memory used by firebase functions when running locally through the emulator? I know it can be done in google cloud (which I have done and can see my functions are working) but im not able to see anything in the documentation and I suspect low mem is causing issues in performance when running locally.
It seems this is not publicly documented and we can assume that this is not currently possible. You can file a Feature Request in the Public Issue Tracker. With every feature request, the Engineering team has more visibility of your needs and they can work on them accordingly to the impact on the users. This is the importance to look first for an existing one. I have searched but not found any, so I think the best way to proceed in this case is to create a new one. Please as much details as possible about how would you like to work this feature and if you found any workaround, so, the community in the PIT can implement it.
I'd like to ask is it smart enough to ship a web service/product without creating any CMS?
I mean in order to ship it faster and get some feedback from users it will be smart from my point of view to ship it w/out CMS and build it as you go but Im afraid from technical point of view - would it create any big problem that would be painful to solve?
Tnx.
Well, it depends on what you expect from the website. Should it be only a one-time web presentation, I think it might be okay to ship it without a CMS. I would also consider it fine if you didn't want to run a database for whatever reason (as most of the CMSs use database).
If the product is meant to be more complicated, it's definitely easier to start with CMS than to hack it in the product later.
However, if you decided for the latter way, you might find CouchCMS interesting. Though I haven't tested it myself, it looks very promising and easy to inject into an existing website.
I wanted to know if anyone has been using AirBnB Rendr and is it stable and ok to use in commercial projects or is it still changing a lot?
I'm developing a website which can run both client and server based, this mean I need to be able to render pages and widgets server and client based.
The server is running Node.js, dust.js and has custom server based code to render the pages and widgets on the server side. I need to pick how to handle it on the client side.
Naturally I want to try and not repeat code, but obviously the client is different I can:
Keep my current page based server rendering and develop custom
client side code.
Use backbone.js on client side and keep my server based code the
same.
Use AirBnB rendr that is based on Node.js and backbone to use the
same code on client and o server. AirBnB Rendr Library
I like the 3rd idea very much, but I'm looking for some input from you guys.
Has anyone used it? any experience with it in terms of stability and/or how often their api changes etc?
I've just started playing around with Rendr. If I ignore the learning curve and oboarding friction, I like it a lot and I plan to write my next large production app using Rendr.
Unfortunately, as bababa listed above, the documentation needs a lot of work. There is an explanation of how Rendr works in its README and the example app's README but beyond that you'll need to source dive in order to figure out how the gears are turning. Currently, there is no forum for questions (other than stack overflow :D) and I've had a hard time figuring out its idioms on my own.
Despite all the struggles, I finally see the light and I'm starting to understand why Rendr is so powerful.
tl;dr - If you're willing to source dive and figure out your own workflow, I would suggest using Rendr. Otherwise, I would recommend going old school by writing a traditional client app with a more mature library. (is it too early to say that? =X)
Well given AirBnb is a successful commercial enterprise, there's some validation that the library works well enough for them. This question is probably best answered by watching their github commit log for breaking changes. Given backbone is 1.0 and essentially stable at this point, rendr will probably quickly stabilize, but honestly your fear of instability is probably unjustified. I think rendr looks compelling and although my current project is using a very similar home-grown solution, I would consider using rendr in a future project or even porting our code to rendr. "Stability" per say is much less important to the web development community compared to other situations like packaged or embedded software.
I used (tried to use) and Rendr on a project and gave up. There are just to many limitations (currently) and the lack of documentation doesn't help. I ended up need to rewrite the source code to accomplish some things I would consider trivial with other frameworks, such as passing multiple collections to a view. It just wasn't possible (at the time I used it) and that was a deal breaker. Not being able to pass a collection of categories and results to a page was to much of a limitation.
I have no doubt it will eventually be ready for production use, but right now I would say unless you are an engineer at AirBnb and know how to hack the source then no, it's not ready.
If you really want to know if it will work for your needs, take a look at the issue list on github. That will give you a good idea where the projects at.
I am trying to create a complete session managment in nodejs for logins, chat sessions etc.
I googled a lot and every solution that i got was with some framework/module. I don't want to use any module/framework. I would rather like to build my own solution for this:
So this is the plan:
I will set a session cookie on the client machine (yet to figure out how)
For each cookie, i will be maintaining a unique id in the database instead of files as is the case with php (i am using mongodb)
When a user opens the application, a cookie will be set, a entry will be made in database and corresponding information from the db will be fetched.
I am yet to lay a concrete plan for this. I wanted to know whether doing it this way is a good idea? i read somewhere....'Real men don't use any framework. They make everything on their own' :P
Please correct me if i am on a wrong direction. M just starting with these things....
I'm not aware of any node.js frameworks that are closed-source. Just pick one that seems to do what you want to do, download it, and study the source code to see how the developer implemented it. Then come up with your (perceived) improvement on how they did it. You'll probably find that implementing session management involves a whole bunch of nitpicky details that were never obvious to you.
Ignore all the above advice if this is a school assignment where you're not allowed to look at related code. If that's the case, I pity you because you have an incompetent teacher.
I would like to add a "tag cloud" to a project I'm working on. I see tons of them via google, but they seem to mostly be "enter an url" type.
Here's an example of what I mean:
I'm looking for one which either has either
a nice web-accessible api
a standalone local executable (linux preferred)
a linkable library (c,python preferred)
of course, other options and suggestions appreciated!
update: it seems what I am looking for is commonly called a tag cloud and not a text cloud, even though I am interested in using it to view blocks of text.
update 2: the Most Excellent Jonathan Feinberg and IBM have release Wordle... hooray!!!
http://www.wordle.net
This question is old and already answered, but I would like to say that Wordcram seems to be very nice. And it's open source.
I'm not sure if you are referring to a simple (ala Flickr) tag cloud, or something a little more complicated like Wordle.
Anyway, if you are looking for a simple tag cloud, it wouldn't be too difficult to implement it yourself (as long as you already have the ability to render HTML) as it is just changing the size and/or colour of each item based on its frequency (or some other measure).
If you want to use an existing library you could look at one of the opensource php versions, like Tag Cloud, put just run them locally on your machine using php rather than through a web server. Just install php and run php filename.php similar to how you would execute a python script.
Looking at the Wordle service, there appears to be no way to automatically create one, as they use a java applet to generate the graphics, which cannot easily be scripted using curl. They do have a question in their FAQ about an API however:
Could you expose Wordle as a web
service that generates images?
A scalable web service should take no
more than a few tens of milliseconds
to do its work. To create a Wordle
requires multiple seconds in a Java
runtime. (That pretty animation is not
for show; it's really laying things
out during the animation). Therefore,
Wordle will always apportion the
CPU-intensive stuff to you, the user,
and your CPU.
As of this writing, Wordle is
sustaining 10 hits per second. There's
no way on Earth to render Wordles at
that speed. Well there is a way, but
it involves way more money than I've
got.
Also, this previous question may help.
Here are two Python-Versions of a tag cloud:
https://github.com/atizo/PyTagCloud
http://peekaboo-vision.blogspot.de/2012/11/a-wordcloud-in-python.html
I search a lot these days and it seems that those two are some of the few "stand-alone" tag cloud generators, which run in Linux (in particular those run in python) on the command line.