I'm super confused to come here and see people complain about its performance.
For me it runs lightning fast, especially compared to other tools like Outlook. Clicking any message loads it instantly, searching through my 30000 emails in 10 different accounts is also instant, etc. Why is my experience so different? Mostly everything is on default settings.
Pretty much the only complaint I have about this tool is that parts of the window sometimes flash for seemingly no reason when it is left open for a while.
> I'm super confused to come here and see people complain about its performance.
I can't complain about performance. My profile folder is ~90 GB. I have multiple email accounts connected and a large local folder which contains email from Google Takeout since I had to delete some (larger) emails from my Google mail to make space. That being said, it would be nice to not see application not responding messages on Fedora (Gnome) as much.
In my experience, sending email in Thunderbird takes an average of 5 seconds or more with just a popup "sending" sitting there. Why not put it in background processes like the other email clients.
I've never seen this implemented in a way that isnt confusing. It goes to some "Outbox" and you no longer understand why it stuck there and how often it's retrying. You can also have weird failure modes where the mail is sent but not saved to the Sent folder. This is also not obvious to display in a background process
I prefer email sitting in the outbox for a short period of time before send/receive completes, it has saved me from mistakes enough times that it is worth the delay.
You haven't used it with imap and lots of mail then. I get 100…500 emails daily and tb freezes for 15…60 seconds at a time. My inbox is kept in order by a bunch of filters, but the thing still freezes if I have lots of e-mails in another mailbox.
The mailboxes are very inefficient, being regular mboxes with an index. I'd rather have the e-mails in a sqlite database. A Maildir on disk is a waste of inodes and a liability when one does a search or archives the thing. Old mail is rarely touched and sqlite has a full text search function.
The interface is okay. Quirky, dated but okay. I just need a snooze option like Gmail's but one that doesn't hide the e-mail. Now I have to convert the e-mail to a task and add an alarm to the task which requires a lot of clicks and setting options that are unergonomic.
>You haven't used it with imap and lots of mail then. I get 100…500 emails daily and tb freezes for 15…60 seconds at a time. My inbox is kept in order by a bunch of filters, but the thing still freezes if I have lots of e-mails in another mailbox.
I do, around 100k mails in various dirs with order of magnitude higher volume than you. Just need to set it up
> The mailboxes are very inefficient, being regular mboxes with an index. I'd rather have the e-mails in a sqlite database. A Maildir on disk is a waste of inodes and a liability when one does a search or archives the thing. Old mail is rarely touched and sqlite has a full text search function.
You can set it to maildir style, sadly impossible after account creation and requires unbelievably much effort for such simple change.
Like, this piece of shit asked me to restart client to change IMAP server name, who wrote that garbage?
Out of curiosity: We're talking "magnitude higher", so that means 1k to 5k mails each day.
How do you cope with that? Given a work day has 8 hours, you'll have at best 30 seconds to process an email (480 working minutes for 1000 mails) and that includes reading, answering and doing the actual work required to have an answer. That seems like not manageable?
On my Linux machine, there's noticeable latency when scrolling through the messages pane. I use the vertical (side by side) layout, but the delay disappears when I use the stacked layout. The jankiness seems to be a function of the messages pane's height.
Outlook is very slow for me, but I only use it on my work laptop.
Thunderbird is almost as slow, and that's on my own desktop.
Even something simple like mousing over emails in the list is slow - there is a noticable delay between the mouse reaching each email, and the grey highlight.
I moved from claws-mail (because O365 now requires oauth and non-thunderbird experience with it is absolutely fucking awful) and it is noticeable slower. Not enough to call it "slow" tho, it is pretty snappy
R2 storage might be within the same order of magnitude as S3, but once you start serving content to users, it becomes many orders of magnitude cheaper.
The average person, when not allowed to use a convenient password manager, will either use the same password for every site or come up with a predictable pattern. Encouraging a password manager helps make sure they don't get destroyed completely when a blog they signed up on 5 years ago is hacked.
This is partly because so many things want an account now. I have over 500 passwords saved, it would be straight up impossible to remember unique strings for each site.
> This is partly because so many things want an account now. I have over 500 passwords saved, it would be straight up impossible to remember unique strings for each site.
I'd argue that it is not just about remembering passwords. A password manager also helps you remember that you even have an account. I have a similar number of passwords stored as you and there's no way I'd remember all the sites I signed up for (never mind the passwords) if not for the password manager I use.
If I was diligent I could probably track them via confirmation emails (or self-authored "confirmations" for services that don't send a confirmation email), but I can guarantee that a lot would slip through the cracks if I were to attempt that.
A password manager also gives you a convenient place to store (and share) secrets, recovery keys, SSH keys, and similar bits of security related information that you cannot memorize.
In addition, password managers are much better than the average human at avoiding phishing attacks. The chances of the password manager auto-filling your normalwebsite.com password on nomalwebsite.com or normal-website.com are infinitesimally low compared to the chances of an average user doing so.
I use BitWarden, but I'm thinking of self-hosting BitWarden. BitWarden's commercial offering might be tempting to crack from a hacker's perspective, but I don't think they would go after a specific user's instance.
there's a non-zero risk of hosting it yourself and not keeping up with the maintenance/security updates of whatever server you host it on. gotta weigh that in the calculus. It might not be likely that they target you specifically but there could be a drive by bot that slurps up your password vault.
Most people seem to think it needs to be accessible online. Remote Access =/= Internet Access. Self hosting an external vault, using VPNs, and requiring MFA access make the vault tricky to get to in the first place. You’re machine would need to be compromised first for an attacker to even connect to it—and at that point you’re compromised (and probably keylogged).
If you’re actively under attack no Password Manager, mental algorithm/ password pattern, Yubikey, or MFA will prevent someone from just using your authenticated session(s).
Does that mean we shouldn’t use these mechanisms? Of course not. When the risk is only realized with full compromise—saying XYZ could pose a threat is moot from a security perspective.
> Self hosting an external vault, using VPNs, and requiring MFA access make the vault tricky to get to in the first place.
ok but that also is prone to a weakness in any part of that chain assuming you even set it up properly in the first place. each piece is another layer that can be hacked or improperly setup.
> This is partly because so many things want an account now. I have over 500 passwords saved, it would be straight up impossible to remember unique strings for each site.
How many of those accounts are essential?
I create throw-away accounts on the regular.
There are many accounts, but few that really matter. For the ones that matter I take care to make sure I have the passwords. For the rest, who cares :)
If you go for the shared strong secret part "uniquefied" by an added per-site trivial part (not saying that you should), you increase risk the longer the trivial part gets: if an attacker somehow determines that your password here is 123hacker456news789 they will easily guess that they can get into your Facebook using 123face456book789. Less easy if it's 123h456n789 (Because you don't really use the spaceballs password, in reality h and n don't stand out half as much). If it's 123c456w789 guessing Facebook's 123c456o789 from that would be quite close to brute forcing unless they get their hands on dozens off that kind.
> This presentation contains absolutely nothing of value for a real programmer trying to actually make stuff.
Speak for yourself! I got a lot from it that feels immediately practical for my current projects as well as others I want to take on, and it makes logic programming concepts and benefits much more accessible to me than any amount of Prolog literature has so far (granted I came in wanting to be compelled).
Do you have some examples of things in here that you think will help you solve current problems?
I'm approaching it from the angle that I don't see how functional programming is at all useful for the kind of gameplay programming that would be done in a "metaverse" or is done in Unreal Engine today. So "it can do functional programming stuff" by itself doesn't cause any excitement. Rather the opposite. There are zero code examples of using it for anything besides abstract math. It executing code out of order seems to serve no purpose besides letting people create undecipherable monstrosities with it.
I'd maybe sum up this presentation as "we took functional programming and tried to make it less unusable by letting you actually mutate state" but I'm still lacking the original motivation of why I'd want to use that in the first place.
Well I don’t build games, so huge lump of salt, but I could benefit from and imagine translating into my work the language’s approach to:
- null safety
- deferred evaluation of expressions with unresolved dependencies
- types as first class values
- the real thing functional programming is about: making it possible to know what value is bound to a thing at any point in a program
The first is valuable for any program, the last is valuable for any program with multiple inputs (so, games, but also basically any human facing program because IO, or really just any program with users and real world interfaces), and types as values is a world of oysters if you like what types afford for development.
Edit: lol I forgot to expand on deferred evaluation, it’s a really good model for distributed users but could also be a good model for highly dynamic applications like one I maintain where there’s a hard constraint on synchronous execution but currently a very lazy model of what needs to be executed.
Logic programming and games actually go well together... inform (probably the most important text adventure engine) is notably a logic programming language. And many game engines have moved to an architecture in which game assets are represented in a database, and changes to game state are represented as queries on the database (which is canonically what logical programming does).
For instance this might be super mario...
mario moves forward or
the gumba moves forward or
mario hits the gumba and mario enters death animation and the gumba stops or
mario hits the gumba from above and the gumba enters death animation or
mario hits a coin box and a his coins increase and mario stops and the box enters coin animation.
For me it runs lightning fast, especially compared to other tools like Outlook. Clicking any message loads it instantly, searching through my 30000 emails in 10 different accounts is also instant, etc. Why is my experience so different? Mostly everything is on default settings.
Pretty much the only complaint I have about this tool is that parts of the window sometimes flash for seemingly no reason when it is left open for a while.