Welcome to my personal blog. If you're looking for my academic profile page, click the UCD in the upper right corner. Or maybe you're looking for my open source development blog?
14 Jan 2019
I made my resolutions on this tweet but it helps to repeat.
How’s it going so far?
20 Jun 2017
Apparently, the Firefox 54 release a week ago was a bigger deal than I had originally thought.
As I was toiling away on my WebExtensions addon, I took note that Firefox 54 was going to enable the sidebar API, so you can pull up a sidebar drawer from your extension. I mentioned last month in my rough plans of porting Perakun to WebExtensions that it’d be nice to have a drawer that holds the list of word lookups just like in the legacy Firefox addon, and theoretically, it is now possible to do that in WebExtensions. (By the way, I forked another Rikai-kun-based addon that looked further developed, called rikaigu, and the addon works in Firefox and Chrome: check out rikaigu.we.) But I also learned in that time that, while there was discussion of adding it to Chrome as well, in early 2016, they decided not to proceed with sidebar API. So looks like if the addon’s going to be cross-browser compatible, I’m going to have to ditch the idea of a sidebar.
But there’s a bigger reason to be excited about Firefox 54 and that’s the multiprocess architecture revamp! I’m looking forward to web development without the browser choking out because I toggled the DOM inspector while a monster of a page is loading…
Except multiprocess functionality is disabled if you make use of any addons that are “multiprocess incompatible”, i.e. utilizing deprecated APIs.
Perapera Japanese being deprecated is not so bad because our Rikaigu port is essentially a drop-in replacement. Woo!
TiddlyFox is a hard one to lose. I’m pretty dependent on Tiddlywiki for jotting notes, and without this, the Tiddlywiki can’t overwrite itself when you press save but prompts you to save the file to disk. You can just point it to the same file and it will function the same way as a workaround, though.
Also looming at the end of this year is Firefox 57 when some more addons are going to stop working because they’re not WebExtensions. I’m really worried about Tree Style Tab possibly going away in November.
So I’m excited about the changes that have been brought on at the expense of breaking API support for all these addons. I feel like there’s some vocal opposition from addon developers who are getting shafted for something they don’t necessarily agree with from an engineering perspective, and I really do feel for them. While I can see that there’s nothing that can be done if the WebExtensions API is incomplete and you just can’t rewrite what you have if it’s not supported, I believe it’s going to be continually improving, and I really hope that some addons end up living into next year.
I’m also looking to join in the fray with Rikaigu, but that progress report will have to wait until another blog post.
One of the questions I had in my last post
was about how to manage sessions and login to a Rails app from a WebExtensions
client. For example, let’s say your Rails app provides a Profiles resource that
you handle in Rails with a route
GET /profiles/:id. Let’s say you provide an
API endpoint as well, at
GET /api/v1/profiles/:id. From your WebExtensions
script, you would issue the XHR like so:
Most likely, the endpoint will be protected by some sort of authentication check.
The nifty thing is if you happen to have cookies from, say, having logged in through
the web interface of the app, those are automatically sent when you send the XHR,
and you can potentially be authenticated that way. But more typically for an API, you might be using
Rails token authentication.
authenticate_or_request_with_http_token, which will look at
the Authorization request header for a token that will be used instead of
cookies. So you’ll need to do two things:
Conventionally, you might see the token made available by the app through some sort of user settings page where the user can copy and paste it into the extension to use for authorization. There might be a more automated way to do this, though.
The second part can be achieved by adding the following to the above snippet:
With this, Rails should be able to pick out the token from the request header and use that to authenticate the request and get your JSON response.
So while you now theoretically have full access to the API, there’s another thing to consider called Cross Origin Resource Sharing (CORS). Basically, it’s a way for a server to specify access control to resources from cross-site scripts, which is what our browser extension is. For example, we don’t want a malicious script to be able to send DELETE requests or otherwise mess with your data. CORS is a way to permit or deny requests, based on where the originating script is coming from.
This may not be an issue because a malicious script would somehow have to get access to the authentication token. But this may mean extra server-side setup depending on how this protocol is enforced across browsers. I’ll investigate the details in the next few days.
13 May 2017
Perakun is a Firefox extension that helps Japanese learners by providing dictionary lookups for mouseover text. A great feature is its ability to save lookups to a word list. This gif gives a good summary:
As I mentioned previously, Chrome has taken the lion’s share of the browser market, and Firefox has plans to move to an addon subsystem based on WebExtensions, which will be compatible with Chrome.
Perakun exists as a legacy addon in Firefox for Japanese and Chinese, but only the Chinese WebExtensions addon exists in Chrome, with only a subset of the functionality.
In theory, an addon written in WebExtensions should plug and play in any supporting browser, but there is enough inconsistency in API support where the code will need some updating.
Currently, I’ve gotten started fixing up the WebExtensions addon, which you can find on github. I’ve made some simple replacements to some deprecated calls, but it still isn’t functional in Firefox.
Things I need to figure out still, are:
document.caretRangeFromPointhas no equivalent function in Firefox but is being used in Chrome to get the mouseover text. It may be the case that the old way of doing it is still compatible and I will have to port it back in.
09 May 2017
I’ve had this idea of hacking additional functionality onto Perakun for a while now, to make it more useful for taking snippets while reading Japanese. Now that I’m actually diving into it, I’ve learned that there’s a lot of doom and gloom in the Firefox extensions community surrounding the new WebExtensions addon system.
Apparently, the old addon system is going to be phased out by Firefox 57 (slated for November 2017) in favor of WebExtensions. Cited reasons for the move include a need to optimize performance by moving away from the old system (i.e. Gecko) and cross-browser compatibility – WebExtensions addons ostensibly plug into Chrome or any compliant browser with minimal effort. Seeing as how Chrome has 60% market share as of this post, it would be nice to write an extension that can reach that many more users. But developers protest, saying that WebExtensions is still a couple of years out from being even feature-equivalent with what the current addon system offers. The future of some big-name addons beyond the end of this year is unclear. Personally, I am worried about what will happen to my Tree Style Tabs and TiddlyFox support.
But technology trends aside, I really want to take this chance to try my hand at modifying a WebExtensions addon. Chinese Perakun is available in Chrome, but conspicuously, the Japanese version is missing, so I have a straightforward goal:
I’ve already made headway with my first goal by setting up the cperakun repository. I’ll post a writeup when I’m able to get it up and running in Firefox.
Computer science grad student at UC Davis, writes C++, enjoys learning languages, and occasionally video games.