conspire | A real-time collaborative editing platform built on Git | Collaboration library

Testing memory leaks with creation of multiple @Dependent instances with WildFly 18.0.1

I am trying to load and process images with an unique crop factor learned from each image. I keep getting an error stating I can't use a tensor as a Python boolean.

For each image, I want to threshold one row of pixels from the center of the image and calculate the percent of pixels over some threshold. I want to use that percentage as a crop factor.

I build a tf.data.Dataset.from_tensor_slices() with version 2.0. My input is a one-dimensional array, which contains indexes for clipping a large numpy array (60 GB).

My Pipeline so far reads the array with np.memmap and should then clips this array. Therefore, I create an array in the dimensions (n, 4), where n is the number of samples. This (n, 4) array is prompted to tf.data.Dataset.from_tensor_slices().

Afterwards I want to call dataset.map(), where the input is one row if the (n, 4) array, which has the shape of [4,]. However, I can not eval the single values of the this tensor, whereas I can evaluate the tensor before the .map() call.

Here is a minimal working example with the error I get:

I am using a yaml file to store some config data, including some (many) regex strings that I don't want to keep in code. Everything works except when I try to search for some incorrectly escaped pipe characters with r'\\\|'. I tried quoted, unquoted and literal strings in yaml, nothing works. Yaml and Python string escape rules together seem to conspire to keep the number of backslashes in a string even. I open and load the file with

I'm transcribing legal documents that come with ordered lists, but there's a catch. Law-makers are lazy and when adding a new clause between clauses 2, and 3, they name it 2A.

As an example, here's some original text:

6 High treason
(1) Any person who abrogates or subverts or suspends or holds in abeyance, or attempts or conspires to abrogate or subvert or suspend or hold in abeyance, the Constitution by use of force or show of force or by any other unconstitutional means shall be guilty of high treason.
(2) Any person aiding or abetting or collaborating the acts mentioned in clause (1) shall likewise be guilty of high treason.
(2A) An act of high treason mentioned in clause (1) or clause (2) shall not be validated by any court including the Supreme Court and a High Court.
(3) Majlis-e-Shoora (Parliament) shall by law provide for the punishment of persons found guilty of high treason.

When coverted to HTML, it's marked up as:

From reading the WSGI specification in PEP 3333, it is not entirely clear to me if it guaranteed that the iterable returned from an application will be iterated on the same thread that run the application or even if all iteration steps are done on the same step (i.e. if the iterable has a thread affinity to the thread that returned it). I found some discussion on the web-sig mailing list from 2005 which says something to the effect of "not guaranteed but it would be bad if some server did it that way".

Also, in the section on middleware, the PEP says that "synchronous applications and servers can conspire to reduce the number of threads that are required to run a given number of application instances simultaneously" which would also imply that iteration may be done from different threads.

So my main question is: Are there any WSGI implementations that iterate the iterable returned from an WSGI application from different threads, esp. threads different from the one that called the application?

One of the comfort benefits of hosting a development version of a web application on the developer's client machine is the ease of which a change can be verified.

In a traditional ASP.NET web application I'll usually have this setup: The IIS site or IIS web application has its physical path pointing to the ASP.NET web application's project path.

This enables me to do a change, recompile if necessary, and hit refresh in the browser to instantly see the effects of my change.

Now, if I wanted to I could have IIS point to a different path and publish the ASP.NET web application to that location with each change, a far more cumbersome process.

I prefer the in-place IIS setup for faster turn-around, and I would like the same setup for ASP.NET Core, but I've not figured out how. As far as I can tell, I can at least do the following:

• Start the debugger in VS (IIS Express or dotnet.exe)
• Run the web application from the console
• Publish the changes to a different physical path that is mapped to by an IIS site/web application.

Both these options work, but circumstances conspire to make the result difficult to manage: I usually work on both the client and the service, and the client frequently involves multiple service web applications. So I will typically need to have multiple web applications running at the same time, and need to update these frequently.

While IIS provides this opportunity in my classic ASP.NET applications, I'm not dead set on continuing to use it. Any other method to achieve smooth turn-around would be acceptable :)

So, any tips?

-S