LiSS | Pytorch code repo for the paper `` Towards Lifelong Self | Machine Learning library

My Rails 6.x app has been using ActiveStorage, attaching various files to my models and it has been working fine. But at some point, the url_for method started producing urls that are unusable (i.e. the server returns a 404 when I try to access the url returned by url_for). The url's are broken in both my dev and prod environments which are using Disk and S3 services as defined in my storage.yml respectively.

To try to figure out the problem, I created a new app with the same versions of Rails (6.0.3.2), aws-sdk, etc. and made a simple model with an attachment, attached in the same way, and everything seems to work fine. So I am guessing that there must be some cfg issue that I missed, but I've not been able to find it.

In both cases, the image file gets created in the location specified in storage.yml and the url's generated both look quite similar, but in the first case; accessing the url produces a 404, but in the new app, it works just fine (i.e. the url returns the expected image)

For reference, I've posted below the code used to create the models and attach images, in case that is of interest.

Any thoughts on what I may have mis-configured or forgotten to migrate appreciated!

Broken app:

I am having some issues with Ruby's OpenURI follow redirect functionality.

When going to a URL that contains %20 in it, and that redirects with a 30x, Ruby's OpenURI fails.

• The exact same URL, with a + instead of %20 works.
• Both the %20 and + versions work properly with curl -L (follow).

Let's say I have the following sentence in HTML:

Please enter your licence number

The screen-reader mis-pronounces the word 'licence' as "liss-ens" (phonetic spelling). It should be pronounced "lice-ens" (phonetic spelling).

I want to fix this by providing a phonetic spelling to the screen-reader, while having the text visually appear the same.

I could use s, aria- attributes and styling as follows:

Please enter your licencelicense number

This works well enough, except that the screen-reader (I'm testing with VoiceOver on MacOS) pauses when it gets to the first span, forcing me to press [VO]+[Right Arrow] to advance to the next word:

"Please enter your" ... [VO]+[Right Arrow] ... "lice-ens" ... [VO]+[Right Arrow] ... "number"

I want the screen-reader to read out the sentence smoothly without pausing.

Is this possible? Or should I not be trying to control this?

I integrated Drupal 8 autocomplete module in my project. Its working fine. At the end search suggestion liss, there is a link for view all results. When I click on the link, it goes to localhost/sampleapp/search/node?keys=test. I got all the results for the keyword "test". But the problem is when the keyword is tes instead of test, I got no results. For example localhost /sampleapp/search/node?keys=tes

I'm using feedparser to parse rss feeds such as https://www.relay.fm/analogue/feed and can't work out how explicitly identify the itunes:category values.

Looking at the feedparser itunes tests it appears that both the itunes:keywords and itunes:category values are put into the feed['tags'] dictionary.

From the tests for category:

I did a plot explaining occurrences of each modality for many variables. It is about clustering problem to show which variables are explaining each cluster. So

We have an array of integer numbers. We want to know for each element whether that element is contained at least in one LIS of many LISs of our array or not. We want to know this for all elements in the array in less than O(n²).

For example array [2, 4, 3, 2, 5] has two LISs. All elements in the array belong to at least one of these LISs, exept the 4^th element which does not belong to any LIS.

I know an easy solution which uses dfs, but its runtime is O(n²).