Hacker News Reader

Comment by newscracker | Hacker News Reader

There are many things planned, especially moving to "web based technologies" for the client as a longer term initiative. The roadmap also has some work on JMAP support, which I hope picks up on mail servers too. One thing I'd love to see, which probably won't be on the roadmap, is native integration with Exchange and Exchange calendaring. I know there are some extensions for this, but none of those have actually worked well for me.

View on HN

A move to anything similar to Electron is gonna be the same to me as the project dying. In either case, I'll no longer use it. That's a shame.

[EDIT] "but it's already XUL!" right but XUL-based GUI programs can run great on machines without enough memory to use Slack or VSCode without hitting swap, even if those were the only programs open. My source for this: I've been using XUL apps since the days when machines with 128MB of system memory were pretty damn high-end. They were kinda heavy back then, but basically fine. They're straight-up lightweight by today's standards.

Thunderbird already needs HTML and CSS for rendering emails, and needs JS for XUL anyways, right? So Thunderbird already needs HTML, CSS, and JS engines at any given time. Where will the memory bloat come from by moving to using HTML, CSS, and JS for the UI?

It's the culture of Electron and NodeJS/NPM and their programming style. Switching to Electron wouldn't require adopting those, but the practices associated with the "modern JS" crowd are almost certain to end up infecting the way that Thunderbird is developed just like it has infected essentially everything about the way anyone does JS these days.

“Web-based technologies” doesn’t necessarily mean something like electron. Firefox used the same language and has largely completed their migration away from XUL. You can see details in the roadmap here: https://docs.google.com/document/d/1ORqed8SW_7fPnPdjfz42RoGf...

I definitely don't consider Firefox's resource use (battery, CPU, memory) something to aim for. Pre 2.0 Firefox? Hell yes, it was part of what got everyone on it in the first place. After 2.0, when it started to bloat rapidly for unclear reasons? Not so much.

Firefox 2 was released 14 years ago. It's perhaps time to upgrade your computer or to stick with old software.

I'm still unsure why certain websites result in browser processes using up substantial resources in terms of CPU time or system memory. I typically will search for "Web Content" processes in top and kill the ones that are taking more than a certain percent of resident memory or using more than 20% CPU time just to keep the computer usable.

That's not a problem I recall encountering in the days of Firefox 2.0.

Firefox 2 had other issues. You don't remember that a website could freeze the whole browser and that it didn't have any thread or process separation between tabs?

I do remember the days where there was no thread or process separation, but, at least for the set of websites I would frequent back in those days, it rarely happened. And, even then, it didn't render the computer unusable due to constant swapping due to insufficient memory.

I think you should really consider to upgrade your hardware if a website can make your computer unusable. I agree it shouldn't be the case but that's how it is.

It really only happens with certain websites, like Facebook. If I leave a browser tab open overnight, then it can get to that point where I see a browser tab taking 50% of the resident memory,

I guess it's too much to ask companies not to implement websites that leak memory.

1) Yes, I know, that's how long Firefox has been going down the wrong path, performance-wise. They've only recently seemed to kinda start addressing that ("now we're as good as Chrome, finally!" OK but Chrome is also bad, so....)

2) I'd prefer software just, you know, not be bad. Safari, for all its problems, demonstrates that there's nothing inherent in a modern desktop web browser that requires it to be grossly inconsiderate of your system resources.

I understand your points but please don't argue for good software and use Safari as a good example. Safari is the new internet explorer, it requires its own workarounds, it doesn't support some standards because Apple doesn't want it, and you must buy a Mac to debug it.

The context was how good Safari is for end users when it comes to resource use, not how good Safari is for developers when it comes to developing for it. These two aren't very related and developers tend to view things HEAVILY biased through their own experience as developers (which is why Electron is so popular).

The point is that Firefox 2 became too bloated for its time and it never felt like it got debloated after that even things progressed.

I remember these days clearly because Firefox 1 was very fast and it displaced Mozilla (what was later called "Mozilla Suite") as the main browser for the Mozilla Foundation exactly because of that performance benefit. However Firefox 2 not only ended up becoming much slower but Mozilla Suite actually became faster and ended up being faster than Firefox despite providing not only a browser, but also a mail client, news client, WYSIWYG editor, IRC client, etc - ie an integrated "internet client".

At the time i liked Mozilla more than Firefox so i was bitter about it and how it was eventually abandoned by Mozilla Foundation (it became Seamonkey, sure, but that is another story).

I remember trying out Firefox for the first time (it was v2.0) and it would take well over 5 minutes before it would even display a window on my computer (regularly, not just the first ever launch). I would have to check Task Manager to make sure that it did start.

IE6, on the other hand, would be usable in less than 10 seconds from launch. Granted, it probably got preloaded by Windows, but even Windows took less than 5 minutes.

> I've been using XUL apps since the days when machines with 128MB of system memory were pretty damn high-end. They were kinda heavy back then, but basically fine. They're straight-up lightweight by today's standards.

Right, but even XUL has grown quite a bit since those days - not to mention that it's pretty much dead at this point. Still, something other than "yet another desktop app built around some descendant of KHTML" would be nice.

There are basically four ways this could go:

1. Thunderbird spearheads a Gecko-based alternative to Electron (following Firefox in deprecating XUL)

2. Thunderbird throws in with the various XUL-preserving Firefox forks to maintain XUL as a Gecko-based alternative to Electron (since the Firefox folks don't seem to have any interest in doing so)

3. Thunderbird switches to some native-widgeted toolkit like Qt or GTK

4. Thunderbird switches to Electron

These are in order of my preference. Option 1 also thankfully seems to be Thunderbird's current direction, though it'd be nice if it could spearhead the necessary documentation and niceties for other applications to use Gecko in desktop apps.

Actually, the Thunderbird team throwing its weight behind Kmail/Contact, a client plagued by UX papercuts and a less than stable foundation (akonadi), but otherwise pleasant to use and portable sounds like an awesome idea!

5. A rewrite in Flutter

That'd just be a subset (or perhaps an emulation) of 3, no?

Flutter is native as much as drawing on a canvas is native. While you could argue that Win32 technically ends up drawing on a canvas at the end, they still expose native widgets. Flutter just straight up draws on said canvas.

While that's true, I've noticed that my installation of Thunderbird has recently started allocating anywhere from 30 to 40% of resident memory (%mem column in top), and it takes a very long time to switch tasks to or from Thunderbird. It's an older laptop with 6 GB of RAM, but I don't recall encountering issues like this on much older machines with just megabytes of system memory.

I have used Thunderbird for many years now, but it's starting to give me some problems, so I'm thinking about finding something else. E.g. it always shows the last folder opened as the windows title when switching between tasks, and filters are not always working as they should. Are there any good alternatives?

There are a number out there that I've heard of but never tried. Years ago, I used to use Seamonkey (which was a continuation of the Mozilla application suite and was the predecessor of Firefox and Thunderbird). I might try it again just to use the mail client :).

Eh, your roses might be a bit glassy :-P.

I used Mozilla (the original) on my Pentium MMX with 128MB of RAM and i remember it being so slow i could literally watch the widgets draw themselves one after another whenever i opened a dialog window.

Oh yeah, Mozilla was heavier than IE, which was heavier than Phoenix/Firebird/pre-2.0-Firefox. Plus FF had a popup blocker built in, which was huge in a way that's hard to appreciate these days. Very quick to start, pages (at least seemed to) load faster, no more popups which were like 99% of ad-annoyances on the web in those days, and tabbed browsing, all free and open-source with no ads in the browser itself (see: Opera). That's what sold it, originally.

why evenn care about JMAP. better support EWS/ActiveSync. JMAP is still in such an early stage and no matter if proprietary or not, there is still no protocol out there that does the same things than EWS/ActiveSync.

That seems like a lot of duplicated effort when Davmail exists. I'd much prefer to see them supporting a decent new protocol than throwing resources at chasing legacy use cases like EWS that's not going to be used by a significant portion of their target market anyways.

I don't mean this to be dismissive, I mean that EWS is a massive moving target that will require a significant allocation of resources that would, IMO, be better spent on other things.

davmail is a huge waste of resources and in 2022 it won't work as well as it did on a server.

Oh no, this means we're in for another Electron app. I don't have anything against Electron, but... Man does it get tedious.

Thunderbird is already using gecko. You can't do an email client for general public without html rendering.

I’m more concerned about the Electron/Chrome browser monopoly (not that Firefox has done itself many favors lately).

Maybe they will pull it off with just gecko, that would be neat.

They are just mirroring Firefox's (and thus Gecko's) own deprecation of XUL in favour of (X)HTML. Firefox's top level windows are now (X)HTML with XUL elements, and XUL specific elements and code paths are gradually being replaced with standard HTML elements or custom elements.

Given that the general public has largely moved on to web interfaces or mobile apps for email access, do we really need to render HTML emails in Thunderbird? I have mine set to display messages as plain text.

I wish it weren't so, but more than half of mails I receive are HTML only. I think it's passed station.

As a technical user, how often is the HTML content actually useful to the content of the message? For some anecdata, I'm a heavy user of Mutt, and I get my HTML mails piped through w3m --dump for display.. I might have to invoke a web browser on the content once every few months?

If they could finally make Faceted Search configurable to show a list of results and make that list matching your normal view settings (e.g. not threaded) that would be awesome. It is really a drag that it is always 5 clicks to get a view where the search results are comprehensible.

Reply on news.ycombinator.com