Skip to contentSkip to conversation

Updating 00-INDEX in Documentation/*

DISCUSSION Henrik Austad 3 participants 4 messages Oct 28, 2013 – Nov 19, 2013

lore
Share & tools
lore
#0 lore raw

Hi Rob, Jiri

If this email goes the wrong place, my sincere apologies, my asbestos underwear can be quickly donned!

I've been reading through some of the docs and one thing that has struct me, is that there is a certain degree of bitrot (textrot?). Almost every category has an outdated 00-INDEX. A couple of indexes has been updated (thanks for applying Jiri!), but more remains.

This is obviously not the most important job to undertake, but it serves as a very convenient starting-point when exploring a given topic.

Hacking away at python shows me that of the 254 subfolders 57 has outdated 00-INDEX, either with missing files, or files that has been removed. (see list below)

Now, I'm quite happy to start fixing this, but what I would like to know before I start:

- is it considered a waste of time? i.e. will it be worth the time
- what is best, a single, big patch, or a series of one patch per directory?
- or should someone(tm) kick whoever caused 00-INDEX to be outdated, to fix it?

thanks!

List of filesystems with outdated index

name of dir: (index_not_folder / in_folder_not_index)

filesystems: (5/17)
filesystems/nfs: (1/2)
m68k: (0/1)
serial: (1/2)
leds: (0/4)
arm: (10/7)
blackfin: (0/2)
ide: (1/1)
fb: (3/3)
networking: (3/12)
scsi: (0/8)
power: (1/2)
filesystems: (5/17)
filesystems/nfs: (1/2)
arm: (10/7)
RCU: (0/1)
mmc: (4/0)
fmc: (4/0)
w1: (2/0)
w1/masters: (1/2)
w1/slaves: (0/1)
timers: (0/3)
devicetree: (5/1)
serial: (1/2)
leds: (0/4)
arm: (10/7)
fb: (3/3)
power: (1/2)
mmc: (4/0)
w1: (2/0)
w1/masters: (1/2)
w1/slaves: (0/1)
powerpc: (4/0)
x86: (0/8)
PCI: (1/1)
isdn: (1/0)
powerpc: (4/0)
laptops: (1/2)
vm: (4/7)
virtual: (3/0)
scheduler: (0/1)
x86: (0/8)
s390: (2/2)
filesystems: (5/17)
filesystems/nfs: (1/2)
arm: (10/7)
serial: (1/2)
leds: (0/4)
arm: (10/7)
fb: (3/3)
power: (1/2)
mmc: (4/0)
w1: (2/0)
w1/masters: (1/2)
w1/slaves: (0/1)
powerpc: (4/0)
x86: (0/8)

Signature · 1 line
Henrik Austad
#1 lore raw
Quoted 13 lines · Henrik Austad

Hi Rob, Jiri

If this email goes the wrong place, my sincere apologies, my asbestos underwear can be quickly donned!

I've been reading through some of the docs and one thing that has struct me, is that there is a certain degree of bitrot (textrot?). Almost every category has an outdated 00-INDEX. A couple of indexes has been updated (thanks for applying Jiri!), but more remains.

This is obviously not the most important job to undertake, but it serves as a very convenient starting-point when exploring a given topic.

Quoted 8 lines

Hacking away at python shows me that of the 254 subfolders 57 has outdated 00-INDEX, either with missing files, or files that has been removed. (see list below)

Now, I'm quite happy to start fixing this, but what I would like to know before I start:

- is it considered a waste of time? i.e. will it be worth the time

That definitely depends purely on you, as you are the sole owner of your own time :)

I'd be happy to apply the fixes if you decide to produce them, as keeping the references broken isn't really nice.

Quoted 2 lines

- what is best, a single, big patch, or a series of one patch per directory?

For this particular thing, I'd prefer single big patch. I can handle the (very unlikely) conflict cases.

Quoted 2 lines

- or should someone(tm) kick whoever caused 00-INDEX to be outdated, to fix it?

That's probably more effort than actually fixing it :)

Thanks,

Signature · 2 lines
Jiri Kosina
SUSE Labs
#2 lore raw
In reply to Henrik Austad
Quoted 4 lines · Henrik Austad

Hi Rob, Jiri
Hacking away at python shows me that of the 254 subfolders 57 has outdated 00-INDEX, either with missing files, or files that has been removed. (see list below)

I'm a bit behind on my email just now. (Started a new job last month, everything else got derailed for a bit.)

Quoted 9 lines

Now, I'm quite happy to start fixing this, but what I would like to know before I start:

- is it considered a waste of time? i.e. will it be worth the time
- what is best, a single, big patch, or a series of one patch per directory?
- or should someone(tm) kick whoever caused 00-INDEX to be outdated, to fix
it?

Go for it.

I've had a script that does this since 2007, but it's in the context of creating html indexes for the kernel.org/doc/Documentation directory, and after the kernel.org breakin they never gave me rsync access back (because it's not built into git and all any server ever does is run git, right?) so I couldn't update it anymore. Then they decided that the Documentation directory should just be a raw checkout from git with no html indexes (because git), and presumably they'll be doing the same for
http://kernel.org/doc/menuconfig any day now (how, I couldn't tell you, but it's apparently no longer my problem)...

So it's kinda hard to get enthused about it these days. I wanted to revive my old push to reorganize the documentation directory (https://lkml.org/lkml/2007/5/22/473 and such) but if the word from on high is that a giant raw pile of unsorted files is the optimal way to organize things, who am I to argue?

Same problem for kernel.org/doc/htmldocs: the ones I did way back when (https://web.archive.org/web/20090327025639/http://www.kernel.org/doc/htmldocs/) had both the "one big html file" version and the lots of little files version and I'd watch the build logs and send patches to get the warnings down, but they took that away and replaced it without even letting me know they were doing that, so I've fallen out of the habit. (My understanding was that the kernel.org guys officially don't care about any data that isn't accessed through git, and a web browser isn't git, so...)

*shrug* I suppose all it's still on the todo list somewhere. Maybe somewhere after reviving http://landley.net/qemu and catching that up to current, but it's in there...

Rob

#3 lore raw
Quoted 15 lines · Rob Landley
Quoted 6 lines · Henrik Austad

Hi Rob, Jiri
Hacking away at python shows me that of the 254 subfolders 57 has outdated
00-INDEX, either with missing files, or files that has been removed. (see
list below)

I'm a bit behind on my email just now. (Started a new job last month, everything else got derailed for a bit.)

Quoted 4 lines

Now, I'm quite happy to start fixing this, but what I would like to know
before I start:

Quoted 8 lines
Quoted 6 lines

- is it considered a waste of time? i.e. will it be worth the time
- what is best, a single, big patch, or a series of one patch per directory?
- or should someone(tm) kick whoever caused 00-INDEX to be outdated, to fix
it?

Go for it.

Ok, I'll work my way through the list then.

Quoted 2 lines

I've had a script that does this since 2007, but it's in the context of creating html indexes for the kernel.org/doc/Documentation directory

Hmm, ok, I was just planning to get the 00-INDEX files up to speed, anything beyond that was not on my plan.

Quoted 29 lines

, and after the kernel.org breakin they never gave me rsync access back (because it's not built into git and all any server ever does is run git, right?) so I couldn't update it anymore. Then they decided that the Documentation directory should just be a raw checkout from git with no html indexes (because git), and presumably they'll be doing the same for
http://kernel.org/doc/menuconfig any day now (how, I couldn't tell you, but it's apparently no longer my problem)...

So it's kinda hard to get enthused about it these days. I wanted to revive my old push to reorganize the documentation directory
(https://lkml.org/lkml/2007/5/22/473 and such) but if the word from on high is that a giant raw pile of unsorted files is the optimal way to organize things, who am I to argue?

Same problem for kernel.org/doc/htmldocs: the ones I did way back when
(https://web.archive.org/web/20090327025639/http://www.kernel.org/doc/htmldocs/)
had both the "one big html file" version and the lots of little files version and I'd watch the build logs and send patches to get the warnings down, but they took that away and replaced it without even letting me know they were doing that, so I've fallen out of the habit. (My understanding was that the kernel.org guys officially don't care about any data that isn't accessed through git, and a web browser isn't git, so...)

*shrug* I suppose all it's still on the todo list somewhere. Maybe somewhere after reviving http://landley.net/qemu and catching that up to current, but it's in there...

Rob

Signature · 1 line
Henrik Austad