3 ms·

Evolution vs planned design in Linux ext4 filesystem

62 points by gioele 15y ago

udp 15y ago
I love it when things like this appear on HN - some insight into the politics of a project that we'd normally never see. Sure does make a change from the kind of crap that keeps hitting the front page nowadays.
casca 15y ago
In any collaborative endeavor, there will always be times that it's necessary for the Decider to pick between competing options. Compromise is a powerful tool but beware of the ATM (1) compromise where 53 bytes were chosen because 48 is the midpoint between 32 and 64 (plus 5 header). 1) https://en.wikipedia.org/wiki/Asynchronous_transfer_mode https://en.wikipedia.org/wiki/Asynchronous_transfer_mode
hxa7241 15y ago
This seems really fundamental in large-scale software engineering, and perhaps in information 'economics' generally. It is about the right use of both design and evolution. You can only design with/for information you have in front of you (that is what design means really). So you should put effort into design only for things you clearly know are wanted. But on the other hand, distribution of designed software is practically free. That should obviously be done freely (ideally) because then the value of the software can be most fully realised. But there is something beside that. It also allows the unpredictable value of the software -- in reuse, adaptation, rewriting, etc. -- to be fully realised too. What design is unable to cover, evolution is given free range to take on. ( http://www.hxa.name/notes/note-hxa7241-20111218T1151Z.html http://www.hxa.name/notes/note-hxa7241-20111218T1151Z.html )
sciurus 15y ago
This was in response to XFS developer Dave Chinner's complaint that: "ext4 is supposed to be primarily the Linux desktop filesystem, yet all I see is people trying to make it something for big, bigger and biggest. Bigalloc, new extent formats, no-journal mode, dioread_nolock, COW snapshots, secure delete, etc. It's a list of features that are omewhat incompatible with each other that are useful to only a handful of vendors or companies. Most have no relevance at all to the uses of the majority of ext4 users." and "do ext4 users really need a new, incompatible, difficult to test on-disk formats to solve problems that most people will never hit on their desktop and server systems before they migrate them to BTRFS?" http://article.gmane.org/gmane.comp.file-systems.ext4/30508 http://article.gmane.org/gmane.comp.file-systems.ext4/30508
koenigdavidmj 15y ago
If I recall correctly, ext4 was a hack to get us limping from ext3 (which was showing its age) to btrfs and filesystems of that ilk.

62 points by gioele 15y ago

udp 15y ago

I love it when things like this appear on HN - some insight into the politics of a project that we'd normally never see. Sure does make a change from the kind of crap that keeps hitting the front page nowadays.

casca 15y ago

In any collaborative endeavor, there will always be times that it's necessary for the Decider to pick between competing options. Compromise is a powerful tool but beware of the ATM (1) compromise where 53 bytes were chosen because 48 is the midpoint between 32 and 64 (plus 5 header). 1) https://en.wikipedia.org/wiki/Asynchronous_transfer_mode https://en.wikipedia.org/wiki/Asynchronous_transfer_mode

hxa7241 15y ago

This seems really fundamental in large-scale software engineering, and perhaps in information 'economics' generally. It is about the right use of both design and evolution. You can only design with/for information you have in front of you (that is what design means really). So you should put effort into design only for things you clearly know are wanted. But on the other hand, distribution of designed software is practically free. That should obviously be done freely (ideally) because then the value of the software can be most fully realised. But there is something beside that. It also allows the unpredictable value of the software -- in reuse, adaptation, rewriting, etc. -- to be fully realised too. What design is unable to cover, evolution is given free range to take on. ( http://www.hxa.name/notes/note-hxa7241-20111218T1151Z.html http://www.hxa.name/notes/note-hxa7241-20111218T1151Z.html )

sciurus 15y ago

This was in response to XFS developer Dave Chinner's complaint that: "ext4 is supposed to be primarily the Linux desktop filesystem, yet all I see is people trying to make it something for big, bigger and biggest. Bigalloc, new extent formats, no-journal mode, dioread_nolock, COW snapshots, secure delete, etc. It's a list of features that are omewhat incompatible with each other that are useful to only a handful of vendors or companies. Most have no relevance at all to the uses of the majority of ext4 users." and "do ext4 users really need a new, incompatible, difficult to test on-disk formats to solve problems that most people will never hit on their desktop and server systems before they migrate them to BTRFS?" http://article.gmane.org/gmane.comp.file-systems.ext4/30508 http://article.gmane.org/gmane.comp.file-systems.ext4/30508

koenigdavidmj 15y ago

If I recall correctly, ext4 was a hack to get us limping from ext3 (which was showing its age) to btrfs and filesystems of that ilk.