Honestly fixing technical debt in companies is not guaranteed to be easy either, or hell getting them on board on any big changes in general. For example, I'm tasked with implementing a message parser for some flight related data... the issue is that the messages are written entirely by hand by a group of people in a specific department. The message type was conceived way back in the 60s, before they even had computers, so humans are expected to read them. As such there's zero consistency in the message format, plus the operatives use red colored text at times to mark a correction, which is lost in transmission by the time it becomes plaintext. It should be obvious to any software engineer worth a damn that this is impossible to parse reliably, so I proposed to write an internal web application designed to make it easier for the operatives to write those messages while it also outputs the messages in a consistent format, making the messages parseable. It got rejected on the grounds of "introducing an interruption in the workflow of the department".
That's the dumbest thing I've ever heard about. I don't know how you didn't have a brain aneurysm pop when you heard that. You should definitely go watch the videos by gar1t starting with "MongoDB Is Web Scale." You'll laugh your butt off.
@@JodyBruchon Oh I've been rewatching gar1t's videos religiously for the past several years. And yeah, I absolutely did have a brain implosion when I realized what cards I was dealt. Eh, I've only just started my professional (read: non-amateur/homelab) IT career, I don't predict I'll last much longer at that company anyway. Just gotta make sure I properly advertise myself as a sysad on the resume for whatever company I end up in next. Dev? No thank you.
I live in windows file manager (Which in and of itself has been getting worse with each revision, I still use Win 7 primarily due to the file manager), the lack of a proper, unified, feature rich file manager is the primary thing that's always kept me from joining team Linux.
I understood the general point in this video, but missed the specific bad decision Linus made in 1995. I backed up a couple of times and think it might be that readdir and related calls don’t return the length of the file name? Maybe I got distracted by the details of the example which wasn’t critical to the point.
That's the crux of the issue, yes. A ton of things that loop on readdir will perform string functions on the d_name. Knowing the length avoids strlen and enables memcpy instead of strcpy. The issue is that the kernel doesn't have a name length in its getdents call structure and adding it breaks the ABI.
The way to fix things without breaking old stuff is by adding a new function/structure. Old applications can keep using use the old one and whomever wants the optimised new way can keep using it, all you need to do is advertise this new function and explain the old one is slower/worse.
Hi jody I watched your latest video and video about that linux is a dead end. My question is from the current operating systems which you use and why? With the latest update upon window 11 that is basically bad for the consumer with the repo feature and Microsoft pushing it user toward window 11 it doesn't feel there are many good alternative. All I want is operating system that I could use for daily things and gaming not worrying company will steal my files and take away my privecy so what is your suggestion because from your are saying all that the current operating systems are all bad. *Just want to be clear that I ask for your gniune opnion because I feel comfuse will the different options of should you move to linux or stay in windows
Already the Linux filesystems are much faster than NTFS, now imagine if the fix that Jody said was to be implemented how much better it would be still. At the end of the day, this is the main problem, the current way is good enough in most situations. It can certainly be fixed, but, yeah, it won't get prioritized unless someone finds a bigger problem from it or if somebody is authoritative enough (like Linus himself) and steps in to push for things like this to happen. While I don't like the guy, I could see Lennard Poettering pushing something like this too. He cares much less about compatibility anyway. It can certainly be done gradually (aka supplying 2 versions, one old one new and keep the old version until most software is also using it) and there's certainly much higher chance of it being implemented like that, but realistically, it would take more than a decade for the old version to get to be removed. I don't remember from the previous rant, Jody, did you try to push or to present the change or the problem with the Linux kernel ? Though I don't know if you can do that so easily, from what I know they communicate using mailinglists and they're certainly busy, so I don't know how feaseable that is. Out of curiosity, how difficult would be to engineer both an "almost worst case scenario" and an environment where you have the fix. That is, patch the kernel, patch glibc or musl or whatever and patch also whatever other programs are used. And then run a test between the current and the ideal situation, to see a difference. Both in speed, maybe energy used (if it can be meaningful) and maybe even lower level stuff, like instructions or cpu cycles used and/or context switches, maybe iops too, though they should be the same. Who knows, maybe a video like that might be impactful enough and it might get traction. Brodie might make a video about it, then the BSD guys will jump in to make fun of how bad Linux is, then the Windows guys will cluelessly embarass themself by saying the same and maybe some guys working in the kernel will be like "ok, I've had enough of this nonsense, let's fix this". I'm a simple man, I like to dream. In the meantime, I'll go and rename all my files and folders into 1 or 2 letter names and also make sure to consolidate as much as possible, to have the least amount of files. For the file manager thingy, I think that Wayland, forcing protocols everywhere, might do/have something for that too. And even without, I wonder if certain apps allow you to specify the file picker program at compile time, or even at run time, like using an env param or something.
Amazon has been working with Canonical for a while to develop Ubuntu, if only to satisfy their own vendetta against Microsoft. They've successfully deployed the OS throughout the majority of the company. For basic business and even basic home use, it is a very viable OS despite whatever tech nerd sees as shortfalls. With all the development that is going into porting apps and games for Linux use, it's a bit short sighted to call the OS dead if all you can do is look at past technical failures.
I think you should look up the concept of the "year of the Linux desktop." It's been around since the late 1990s. We're 25 years into development. It's not gonna happen. 25 years ago we were running Windows 98.
@@JodyBruchon I've been a Linux user since 1994 having used many different distributions as my daily driver. The problem with Linux development is the funding behind it. Red Hat has had corporate dollars behind it for decades, but that OS isn't for home users. RedHat Workstation was intended for business use and as far as Linux goes, it's very well polished for that purpose. This is the same for just about any distribution of Linux that has corporate dollar backing, even Ubuntu, which ostensibly is more ready for home use than any other distro, but with it being backed by Amazon, it's for business use. Windows, on the other hand, has always been developed for both home users AND business users. There have been billions of dollars more going into Windows development than Linux development. Further complicating things are the sheer numbers of Linux distros out there with everyone thinking "I can do it better" and coming up with yet another competing OS. There's an XKCD comic describing this very phenomenon. Right now the actual best home use case comes from BSD and a quaint little company called Apple. Too bad it's a closed platform driven by greed. At this juncture, development time is irrelevant when there aren't any major players developing for home use.
Amazon is also switching from Android to Linux, not to improve the customers' experience, but lock it down even more. Amazon's Fire OS tablets already come with loads of bloatware though thanks to the XDA community it can be removed and improved. I won't be buying their new budget tablets if they don't run Android anymore.
@@PhilipMarcYT imagine spending $100 on a tablet only to discover that it comes with ads literally everywhere and you have to pay a monthly subscription fee to get rid of them.
So why don't you put a group together, and develop a Linux distro that addresses the issues you raise? As things stand, I'd prefer to muddle through with a dual-boot system, using WIN only when I have no choice. What I'd like to see (and I know suggestions are "a dime a dozen") is a video to the effect "Here's how we'll make Linux the future", NOT just Linux isn't. Also, you could address the many inadequacies of WIN. It's clunky , "you're just the piece of garbage who paid for this junk, what do YOU matter" . It seems to me it's a question of WHICH set of drawbacks one decides to live with, and I prefer doing whatever it takes to knock Silicon Valley down more than a few notches.
Because changing even just the API to fix accumulated technical debt is NOT like creating another Linux distribution, it's creating an INCOMPATIBLE derivative kernel: you now have to change ALL the software that uses the API portion that was changed. To some extent it can be automated (e.g. with Coccinelle and its Semantic Patch Language) BUT the remaining changes that (at least with current available tools) cannot be automated, if any, is not guaranteed to require a low amount of labor.
@@s.b.8704 I gathered as much from the videos, and that was one reason I suggested a group to do the work. The product, if it was ever produced, might not even qualify as a "distro", it would have to be so much modified.
as long as in Linux you have the full control of copies on top of free and opensource, is impossible to be dead end. one thing is certain is that there aren't sufficient software and components to complete professional works, holding back Linux from being the future.
Honestly fixing technical debt in companies is not guaranteed to be easy either, or hell getting them on board on any big changes in general. For example, I'm tasked with implementing a message parser for some flight related data... the issue is that the messages are written entirely by hand by a group of people in a specific department. The message type was conceived way back in the 60s, before they even had computers, so humans are expected to read them. As such there's zero consistency in the message format, plus the operatives use red colored text at times to mark a correction, which is lost in transmission by the time it becomes plaintext.
It should be obvious to any software engineer worth a damn that this is impossible to parse reliably, so I proposed to write an internal web application designed to make it easier for the operatives to write those messages while it also outputs the messages in a consistent format, making the messages parseable. It got rejected on the grounds of "introducing an interruption in the workflow of the department".
That's the dumbest thing I've ever heard about. I don't know how you didn't have a brain aneurysm pop when you heard that. You should definitely go watch the videos by gar1t starting with "MongoDB Is Web Scale." You'll laugh your butt off.
@@JodyBruchon Oh I've been rewatching gar1t's videos religiously for the past several years. And yeah, I absolutely did have a brain implosion when I realized what cards I was dealt. Eh, I've only just started my professional (read: non-amateur/homelab) IT career, I don't predict I'll last much longer at that company anyway. Just gotta make sure I properly advertise myself as a sysad on the resume for whatever company I end up in next. Dev? No thank you.
@@Spaztron64 Being someone else's code monkey is rarely fun unless the person in charge is not a middle manager sort.
I live in windows file manager (Which in and of itself has been getting worse with each revision, I still use Win 7 primarily due to the file manager), the lack of a proper, unified, feature rich file manager is the primary thing that's always kept me from joining team Linux.
I understood the general point in this video, but missed the specific bad decision Linus made in 1995. I backed up a couple of times and think it might be that readdir and related calls don’t return the length of the file name? Maybe I got distracted by the details of the example which wasn’t critical to the point.
That's the crux of the issue, yes. A ton of things that loop on readdir will perform string functions on the d_name. Knowing the length avoids strlen and enables memcpy instead of strcpy. The issue is that the kernel doesn't have a name length in its getdents call structure and adding it breaks the ABI.
@@JodyBruchon👍🏻
One more clarification: Jody is a gnome. Support Jody, and you’ll be supporting GNOME.
What would Jody need to be a -KDE- kde ? Have 10 arms ?
The way to fix things without breaking old stuff is by adding a new function/structure. Old applications can keep using use the old one and whomever wants the optimised new way can keep using it, all you need to do is advertise this new function and explain the old one is slower/worse.
What is your car, you seem floating?
But I learned something from this chat anyway.
Mitsubishi Mirage.
@@JodyBruchon Not Lancer EVO with a busty anime girl livery? Lame.
Hi jody I watched your latest video and video about that linux is a dead end. My question is from the current operating systems which you use and why? With the latest update upon window 11 that is basically bad for the consumer with the repo feature and Microsoft pushing it user toward window 11 it doesn't feel there are many good alternative. All I want is operating system that I could use for daily things and gaming not worrying company will steal my files and take away my privecy so what is your suggestion because from your are saying all that the current operating systems are all bad.
*Just want to be clear that I ask for your gniune opnion because I feel comfuse will the different options of should you move to linux or stay in windows
Already the Linux filesystems are much faster than NTFS, now imagine if the fix that Jody said was to be implemented how much better it would be still.
At the end of the day, this is the main problem, the current way is good enough in most situations. It can certainly be fixed, but, yeah, it won't get prioritized unless someone finds a bigger problem from it or if somebody is authoritative enough (like Linus himself) and steps in to push for things like this to happen. While I don't like the guy, I could see Lennard Poettering pushing something like this too. He cares much less about compatibility anyway.
It can certainly be done gradually (aka supplying 2 versions, one old one new and keep the old version until most software is also using it) and there's certainly much higher chance of it being implemented like that, but realistically, it would take more than a decade for the old version to get to be removed.
I don't remember from the previous rant, Jody, did you try to push or to present the change or the problem with the Linux kernel ? Though I don't know if you can do that so easily, from what I know they communicate using mailinglists and they're certainly busy, so I don't know how feaseable that is.
Out of curiosity, how difficult would be to engineer both an "almost worst case scenario" and an environment where you have the fix. That is, patch the kernel, patch glibc or musl or whatever and patch also whatever other programs are used. And then run a test between the current and the ideal situation, to see a difference. Both in speed, maybe energy used (if it can be meaningful) and maybe even lower level stuff, like instructions or cpu cycles used and/or context switches, maybe iops too, though they should be the same. Who knows, maybe a video like that might be impactful enough and it might get traction. Brodie might make a video about it, then the BSD guys will jump in to make fun of how bad Linux is, then the Windows guys will cluelessly embarass themself by saying the same and maybe some guys working in the kernel will be like "ok, I've had enough of this nonsense, let's fix this". I'm a simple man, I like to dream. In the meantime, I'll go and rename all my files and folders into 1 or 2 letter names and also make sure to consolidate as much as possible, to have the least amount of files.
For the file manager thingy, I think that Wayland, forcing protocols everywhere, might do/have something for that too. And even without, I wonder if certain apps allow you to specify the file picker program at compile time, or even at run time, like using an env param or something.
Amazon has been working with Canonical for a while to develop Ubuntu, if only to satisfy their own vendetta against Microsoft. They've successfully deployed the OS throughout the majority of the company. For basic business and even basic home use, it is a very viable OS despite whatever tech nerd sees as shortfalls. With all the development that is going into porting apps and games for Linux use, it's a bit short sighted to call the OS dead if all you can do is look at past technical failures.
I think you should look up the concept of the "year of the Linux desktop." It's been around since the late 1990s. We're 25 years into development. It's not gonna happen. 25 years ago we were running Windows 98.
@@JodyBruchon I've been a Linux user since 1994 having used many different distributions as my daily driver. The problem with Linux development is the funding behind it. Red Hat has had corporate dollars behind it for decades, but that OS isn't for home users. RedHat Workstation was intended for business use and as far as Linux goes, it's very well polished for that purpose. This is the same for just about any distribution of Linux that has corporate dollar backing, even Ubuntu, which ostensibly is more ready for home use than any other distro, but with it being backed by Amazon, it's for business use.
Windows, on the other hand, has always been developed for both home users AND business users. There have been billions of dollars more going into Windows development than Linux development. Further complicating things are the sheer numbers of Linux distros out there with everyone thinking "I can do it better" and coming up with yet another competing OS. There's an XKCD comic describing this very phenomenon.
Right now the actual best home use case comes from BSD and a quaint little company called Apple. Too bad it's a closed platform driven by greed.
At this juncture, development time is irrelevant when there aren't any major players developing for home use.
Amazon is also switching from Android to Linux, not to improve the customers' experience, but lock it down even more.
Amazon's Fire OS tablets already come with loads of bloatware though thanks to the XDA community it can be removed and improved. I won't be buying their new budget tablets if they don't run Android anymore.
@@PhilipMarcYT imagine spending $100 on a tablet only to discover that it comes with ads literally everywhere and you have to pay a monthly subscription fee to get rid of them.
@@JodyBruchon And of course RUclips ate my lengthy reply to the earlier comment. Sigh.
So why don't you put a group together, and develop a Linux distro that addresses the issues you raise?
As things stand, I'd prefer to muddle through with a dual-boot system, using WIN only when I have no choice.
What I'd like to see (and I know suggestions are "a dime a dozen") is a video to the effect "Here's how we'll make Linux the future", NOT just Linux isn't.
Also, you could address the many inadequacies of WIN. It's clunky , "you're just the piece of garbage who paid for this junk, what do YOU matter" .
It seems to me it's a question of WHICH set of drawbacks one decides to live with, and I prefer doing whatever it takes to knock Silicon Valley down more than a few notches.
Because changing even just the API to fix accumulated technical debt is NOT like creating another Linux distribution, it's creating an INCOMPATIBLE derivative kernel: you now have to change ALL the software that uses the API portion that was changed. To some extent it can be automated (e.g. with Coccinelle and its Semantic Patch Language) BUT the remaining changes that (at least with current available tools) cannot be automated, if any, is not guaranteed to require a low amount of labor.
@@s.b.8704 I gathered as much from the videos, and that was one reason I suggested a group to do the work.
The product, if it was ever produced, might not even qualify as a "distro", it would have to be so much modified.
@@occhamite sorry, I tried to reply twice but RUclips keeps deleting my comment, which doesn't contain anything offensive so I don't understand why.
as long as in Linux you have the full control of copies on top of free and opensource, is impossible to be dead end. one thing is certain is that there aren't sufficient software and components to complete professional works, holding back Linux from being the future.
Time to diversify my RUclips content... buh bye...
You don't have to announce your departure. I'm not running an airport!