fix issue #124

[lookfirst]

No description provided.

[plibither8]

Woah, thanks so much for working on this issue! Taking a look...

[lookfirst]

No problem. It is a hacky solution, but it is what it is.

What I'd do is add a cache somewhere so that common selectors like this aren't done multiple times in multiple places... I assume the browser has this cache internally, but it would still be good to add a layer above it for further optimization.

[plibither8]

We can look into adding a session-cache that is available for the current page and refreshes on page-change or reload. I believe persisting such a cache over different page loads can cause problems when the underlying HTML (i.e. the items/comments) change. What do you think?

[lookfirst]

Just a simple in memory LRU cache that flushes on reload.

But, this is a larger architectural issue. I would have designed things a bit differently from the start. Primarily, I'd have a wrapper around document.querySelector() so that I'd have more control over errors as well. Instead of just breaking the entire page and a console message. I think it would be nicer to catch null responses and deal with them more elegantly (ie: a nice message to the user). The caching layer could be embedded in that too.

If/when this thing converts to typescript, I'd rewrite large portions from the ground up.

[lookfirst]

Here is another thing I was just thinking about... think of the page as a database full of data that you want to action on. You'd want to effectively have a data layer that you work against. Right now, you're just plucking what you need from the dom, but instead... map the whole page into an object structure and then you can work against that API instead of just randomly finding stuff in the page.

[lookfirst]

Then, instead of a cache on individual queries, you just CRUD portions of your data structure.

[plibither8]

You just penned down my exact thoughts, pretty much what I was thinking of doing since a long time. As you mentioned, the extension requires a pretty large refactor, and that requires time. My semester at uni will be ending around mid-December, so hope to work on this then! Will definitely loop you in the architectural decisions, and if you can and are willing to, we can work on it together. 😄

Until then, I'll take a look at this PR and move forward with it. Thanks again!

[lookfirst]

@plibither8 ...

[lookfirst]

@plibither8 ... adding some more fixes...