Disclaimer: This is a personal web page. Contents written here do not represent the position of my employer.
Saturday, December 18, 2021
geewallet 0.4.300.0 released!
10th of my 21-day quarantine*! And to celebrate, I'm going to release a new version of geewallet. It's not that I blog about geewallet releases often (or blog at all, lately), but this one is a special one for me. We decided to call it 0.4.300.0
- We fixed the GTK theme for our snap package. (Long version of the story: ever since we upgraded our snap generation process to take place in Ubuntu 20.04 instead of Ubuntu 18.04, the theme stopped working so the app was not showing anymore with the default theme of the system, but with the default Gtk theme, which is very plain. Even if you might consider this issue important, we haven't had time to look at it because we've been very busy finishing Lightning support. Sorry.)
- The chart rendering doesn't use SkiaSharp anymore, but good-old Cairo. This fixes some UI glitches that we had in the GTK frontend. (Long version: for this, we didn't just draw the chart using Cairo in our Gtk frontend, we actually wrote an implementation of the Shapes API for the Xamarin.Forms' GTK backend, and we contributed the work upstream: https://github.com/xamarin/Xamarin.Forms/pull/14235 . Hopefully they merge it soon so that we don't need to use our own forked repo/nuget anymore.)
- Fixed a crash when pairing with a cold-storage wallet. (Long version: user might not know that pairing is only allowed against another geewallet instance; low-hanging fruit bugfix which I shouldn't have neglected for so long, I know.)
- Fixed a crash when scanning some QR-codes that contained unknown parameters in the bitcoin URI. (Long version: I was actually in El Salvador and when trying to use a BTM, I found this bug! Apparently some BTMs here add an extraneous "chivo" param in the URI's querystring, in case the wallet being used is the one from the government; not sure why. In this case, geewallet was failing fast instead of ignoring the unexpected intruder.)
- Our CI now checks that our Android, macOS, and iOS frontends don't break. Previously the only frontends that we built in CI were the Gtk one (Linux) and the Console one (cross-platform, it's just terminal-based).
- We do snap package generation in GitLab now instead of GitHub. This is good because Microsoft keeps changing the Linux VMs being used in the GitHubActions service so we cannot keep up fixing things that just break out of the blue (so, they break independently from what we change in our commits, which is very confusing!). (Long version: we had to use GitHubActions because GitLabCI uses docker under the hood; so given that snapcraft uses systemd, it conflicts with it; now we use a "docker in docker" approach to be able to run in GitLabCI; which also allows us to publish the snap package as an artifact in the GitLabCI pipeline, not just publishing it to the Snap Store; this way, in case you somehow need a previous version in the future you can grab it from there, something that you couldn't just via snap AFAIU).
- Even though this wallet supports two ETH currencies (ETH itself, and DAI), we don't recommend their use at the moment because of the high fees and long confirmation waits these days. This is because the wallet waits for an ETH transaction to be mined (to make sure it didn't run out of gas, and if it did, report the problem to the user), but these days this wait is longer than the time-out. The short-term fix for this is either a) assume it will never ran out of gas, since our address is not a contract anyway (so I guess it can never run out of gas, right? feel free to prove me wrong, my ETH knowledge is not top-notch), or b) have some UI indicating that a transaction has been sent but not accepted by the network yet. The long-term fix is to have off-chain (Layer2) technology supported by the wallet, but we don't know which technology we will choose for this, and of course we're giving priority to the first Layer2 technology: Lightning (which is only compatible with BTC and LTC). All this aside, the wallet works well with ETC (an Ethereum-compatible technology). Anyway, this doesn't worry me too much because... what is the ETH blockchain used for these days, mainly? NFTs and DeFi pyramid schemes. In case you didn't get the memo, most of the former (if not all) are scams, and the latter are all of them mainly based on dubious centralized stablecoins (which could suffer fractional reserve and therefore cause bank runs, as Elizabeth Warren has already warned about).
- Despite this wallet being implemented with .NET (F#), our Windows compatibility story is very poor :'-( We ran into limitations of the Microsoft's AOT technology being used for UWP apps (required by the official process required to publish it in the WindowsStore) in the past. Nowadays apparently you can publish apps in the WindowsStore without these limitations, but we haven't tried again. Maybe by the next time we give it another go, we might have moved to MAUI already (which means WinUI instead of UWP under the hood). As always, if this is your cup of tea, we accept MRs!