EXT4 Brings New Debugging Ioctls For Linux 5.4

[phoronix]

Phoronix: EXT4 Brings New Debugging Ioctls For Linux 5.4

Like the mostly mundane Btrfs and XFS changes with the Linux 5.4 merge window, the EXT4 file-system activity is mostly focused on fixes too but also new debugging ioctls...

http://www.phoronix.com/scan.php?page=news_item&px=EXT4-Linux-5.4-Changes

[cynic]

wait, what? does anything existed before 1-1-1970?

[L_A_G]

I can understand that people wanting to be thorough would want to have support for timestamps earlier than the Unix equivalent of the big bang, midnight on the 1st of January 1970, but what practical use cases are there for this?

Oh and before anyone suggest documenting things from before that, we're talking about something used for file created, modified, last accessed, etc. timestamps. I highly doubt many people have actual files from the days of the Apollo program and before just lying around on their HDD.

[skeevy420]

I can understand that people wanting to be thorough would want to have support for timestamps earlier than the Unix equivalent of the big bang, midnight on the 1st of January 1970, but what practical use cases are there for this?

Oh and before anyone suggest documenting things from before that, we're talking about something used for file created, modified, last accessed, etc. timestamps. I highly doubt many people have actual files from the days of the Apollo program and before just lying around on their HDD.

Birth records, government & judicial records, stuff the Library of Congress stores, the Internet Archive might need it...

Just because 99% of home users don't need it doesn't mean there aren't practical uses for it.

[L_A_G]

Birth records, government & judicial records, stuff the Library of Congress stores, the Internet Archive might need it...

Just because 99% of home users don't need it doesn't mean there aren't practical uses for it.

Back in the 1960s and before those were still paper records and didn't start being digitized until the 1970s, so they don't really apply here. The internet also didn't exist until the 90s so that doesn't apply either and the Library of Congress in the U.S didn't really start digitizing either until well into the 1970s.

Sure, the U.S and many other countries have used computers in government for things like calculating peoples' taxes and doing census since the 1950s, but the computers back then were used just to speed up processing, not storing the vast amounts of information necessary as digital storage was still too expensive for it.

[skeevy420]

Back in the 1960s and before those were still paper records and didn't start being digitized until the 1970s, so they don't really apply here. The internet also didn't exist until the 90s so that doesn't apply either and the Library of Congress in the U.S didn't really start digitizing either until well into the 1970s.

Sure, the U.S and many other countries have used computers in government for things like calculating peoples' taxes and doing census since the 1950s, but the computers back then were used just to speed up processing, not storing the vast amounts of information necessary as digital storage was still too expensive for it.

I dunno, maybe they want to be able to backdate stuff to make searching by date easier or to be thorough and complete. I can see wanting to timestamp something from the 60s with a date from the 60s for that kind of purpose. Yeah, there's databases and etc that can do that too, but that doesn't change that some would prefer to just use a file's metadata to get that information since that's one less layer of complexity to manage.

Like I said, 99% of us just do not need that feature. For those that do need it, cool, now they have it. For those that don't need it, this strays into the "STFU; this doesn't really effect us or our desktops so why are we complaining about it?" territory.

[SystemCrasher]

Treason uncloaked! So now we know Linux plans to stay around at least until 24th century. Whoa.

[L_A_G]

I dunno, maybe they want to be able to backdate stuff to make searching by date easier or to be thorough and complete. I can see wanting to timestamp something from the 60s with a date from the 60s for that kind of purpose. Yeah, there's databases and etc that can do that too, but that doesn't change that some would prefer to just use a file's metadata to get that information since that's one less layer of complexity to manage.

Sure, that's one less layer of complexity to handle those with an extremely niche use case, but a whole bunch of unnecessary complexity/hacks (depending on how it's implemented) to something used by a massively larger userbase.

Like I said, 99% of us just do not need that feature. For those that do need it, cool, now they have it. For those that don't need it, this strays into the "STFU; this doesn't really effect us or our desktops so why are we complaining about it?" territory.

As I said, it's just unnecessary complexity to a codebase used by a very large userbase just for the sake of fraction of that so small we're actually talking about a theoretical one here, not one we actually know exists for sure.

I may sound like I'm being unnecessarily negative towards feature work here, but the open source world (and software in general) has a nasty tendency towards feature creep at the expense of both maintainability and resources available for it. There's probably no better example of this than the absolute behemoth that is SystemD has grown into after initially being intended as a much needed replacement for init by way of launchD from MacOS.