I gave this talk at the South East Linux Fest in 2014. It has finally been put online. I'd really like everyones input.
https://www.youtube.com/watch?v=ymBQPBEVgQI
I am sure that I got some things wrong, so I'd really like if people could point out where I got things wrong... or what things I should add in.
I've been asked to give the talk again this year, and Id like to have input from the puppy community on how to make the talk better.
Thanks for your input. Criticism is welcome.
My talk about Puppy last year is online - want your input.
-
Bindee
Not had time to watch it yet but from what i can see from the first couple of Mins.
You need to grow your hair and goaty longer like Richard Rawlings.
Swigging a beer while wearing a T-shirt that hates Microsoft.
Have Booth Babes in penguin T-shirts as part of the opening before your presentation.
How that for a critique?
You need to grow your hair and goaty longer like Richard Rawlings.
Swigging a beer while wearing a T-shirt that hates Microsoft.
Have Booth Babes in penguin T-shirts as part of the opening before your presentation.
How that for a critique?
My thoughts.
1. Too much is made about puppy running as root or not being a multi-user system.
2. Perhaps some disclaimers should be given about drag and drop for save files.
3. Puppy distributions aren't only built from the binaries of other distributions. For instance I believe that April64 is built completely from source using the T2 toolchain.
Regarding 1, it is easy to make a script so that an application starts as spot. For instance on my system I have the following script to start freenet:
Also the fact that some system is multi-user should not be a selling point if the multi-user aspects of the system are poorly designed. For instance, on windows 8.1 it is difficult to install a piece of software if you are not logged in as Administrator. Trying to run a program as administrator when not logged in as administrator doesn't seem to work well any more. Also the sandboxing between programs in apdata on windows 8.1 makes it difficult for the programs to work together. Finally, anything installed on windows 8.1 under apdata is very hard to access from another user account. I posted a rant about this on sone (A social network plugin for freenet)
Regarding 2, when I was using Lupu (an older version of puppy linux), I had some problems copying save files (Especially between medium that had different file systems ntfs vs ext3 as an example). I think the problem was when the save file was active, copying caused corruption and led to kernal panic. I was advised that one should boot without (or from a different) save file then the one one wishes to duplicate as save file. Perhaps in newer versions of puppy this isn't a problem.
===================
besides all that, great talk, I really enjoyed it
1. Too much is made about puppy running as root or not being a multi-user system.
2. Perhaps some disclaimers should be given about drag and drop for save files.
3. Puppy distributions aren't only built from the binaries of other distributions. For instance I believe that April64 is built completely from source using the T2 toolchain.
Regarding 1, it is easy to make a script so that an application starts as spot. For instance on my system I have the following script to start freenet:
Code: Select all
#!/bin/sh
cd /root/spot/Freenet/
nohup su -c "sh run.sh start" spot&
Regarding 2, when I was using Lupu (an older version of puppy linux), I had some problems copying save files (Especially between medium that had different file systems ntfs vs ext3 as an example). I think the problem was when the save file was active, copying caused corruption and led to kernal panic. I was advised that one should boot without (or from a different) save file then the one one wishes to duplicate as save file. Perhaps in newer versions of puppy this isn't a problem.
===================
besides all that, great talk, I really enjoyed it
That's a nice presentation, very well done. Just a few thoughts:
1. Your talk assumes a frugal installation, understandably. It might be worth mentioning in passing that a more traditional full install is possible.
2. You mention Debian and Slacko as binary bases from which Puppies are built. I didn't hear you mention Ubuntu, although perhaps that was shown on one of your slides. Anyway, it was a little surprising not to hear you explicitly mention the existence of Ubuntu-based Puppies.
3. I don't think your talk conveyed a sense of just how many Puppy derivatives there are, even just counting the ones that are fairly actively supported. If your audience knew this, they would have a better appreciation for the challenges faced in trying to put Puppy development on a more organized footing.
4. I was surprised that nobody asked you what happens if there isn't actually enough RAM to hold everything. Admittedly there aren't a lot of crazies like me running Linux on Pentium 2 boxes...
5. You mentioned some of the efforts to address the "Puppy runs as root" issue. Granted that it's a matter of opinion, and this is your talk, not mine, but I would have concluded with a cautionary "but if you really need a true multi-user Linux, then Puppy is not for you."
1. Your talk assumes a frugal installation, understandably. It might be worth mentioning in passing that a more traditional full install is possible.
2. You mention Debian and Slacko as binary bases from which Puppies are built. I didn't hear you mention Ubuntu, although perhaps that was shown on one of your slides. Anyway, it was a little surprising not to hear you explicitly mention the existence of Ubuntu-based Puppies.
3. I don't think your talk conveyed a sense of just how many Puppy derivatives there are, even just counting the ones that are fairly actively supported. If your audience knew this, they would have a better appreciation for the challenges faced in trying to put Puppy development on a more organized footing.
4. I was surprised that nobody asked you what happens if there isn't actually enough RAM to hold everything. Admittedly there aren't a lot of crazies like me running Linux on Pentium 2 boxes...
5. You mentioned some of the efforts to address the "Puppy runs as root" issue. Granted that it's a matter of opinion, and this is your talk, not mine, but I would have concluded with a cautionary "but if you really need a true multi-user Linux, then Puppy is not for you."
Related to this I'm curious as to how the layered file system works with a full install or is there no layered file system in a full install? Finally, what about a save file vs a save directory. Does a save directory use a layered file system?6502coder wrote:That's a nice presentation, very well done. Just a few thoughts:
1. Your talk assumes a frugal installation, understandably. It might be worth mentioning in passing that a more traditional full install is possible.
It also occurred to me that an interesting application of a layered file system might be to capture all the files that changed after an installation process so that the changed files can be copied into a directory and made into a pet.
Those are interesting questions, but perhaps too much detail for an introductory talk. Although it would be wise to have the answers in hand in case somebody in the audience asks...s243a wrote: Related to this I'm curious as to how the layered file system works with a full install or is there no layered file system in a full install? Finally, what about a save file vs a save directory. Does a save directory use a layered file system?
Hello, Q5sys.
Congratulations are certainly in order for a well-structured talk.
You really make clear that the Puppy difference is layers. I like that
perspective. I knew it intellectually, but you explained it from a more
practical perspective. Thanks for a better understanding.
It certainly accounts for Puppy's versatility and efficiency compared to
the "mammoth" distros and even some of the smaller ones.
In a nut-shell: they move as one block, i.e. clumsily, whereas Puppy
moves as blocks.
As to improving your talk, I'm not sure that it needs any, really.
* As mentioned above, maybe insist a little more on the fact that we're
not parasiting or bootlegging other Linuxes. We can compile our own
stuff, but indeed why recompile a package that already does a good job.
* Maybe mention that full-distro, production-type, Puppies do exist, for
people who don't want to hunt for packages to populate their Puppy.
(The latest in that lineage being stemcee's emSee-tahr-6.02.)
Keep up the good work! BFN.
musher0
Congratulations are certainly in order for a well-structured talk.
You really make clear that the Puppy difference is layers. I like that
perspective. I knew it intellectually, but you explained it from a more
practical perspective. Thanks for a better understanding.
It certainly accounts for Puppy's versatility and efficiency compared to
the "mammoth" distros and even some of the smaller ones.
In a nut-shell: they move as one block, i.e. clumsily, whereas Puppy
moves as blocks.
As to improving your talk, I'm not sure that it needs any, really.
* As mentioned above, maybe insist a little more on the fact that we're
not parasiting or bootlegging other Linuxes. We can compile our own
stuff, but indeed why recompile a package that already does a good job.
* Maybe mention that full-distro, production-type, Puppies do exist, for
people who don't want to hunt for packages to populate their Puppy.
(The latest in that lineage being stemcee's emSee-tahr-6.02.)
Keep up the good work! BFN.
musher0
musher0
~~~~~~~~~~
"You want it darker? We kill the flame." (L. Cohen)
~~~~~~~~~~
"You want it darker? We kill the flame." (L. Cohen)
So far I have just had a quick flick through parts of the vid and it looks very interesting so I will come back to it and fill in the gaps a bit later, but one of the "layering" issues I would like to hear more about in future talks is concerning possible methods for running Puppy via fully readonly sfs in RAM, such that there is no RW savefile at all. (a concept which I consider to be a significant contributer to security and consistency / reproducibility from boot to boot)
As far as I can tell the Puppy layering setup lends itself to adding a personalised sfs over the top of the main sfs, although that is not the traditional puppy way - normally it is the personalised rw savefile that populates the top layer, but I am keen to see other ways of running that personalised layer as an sfs (readonly) for the sake of system stability.
Don't know if you ever try to run your pups that way but it is what I do with my daily driver pup and the concept of readonly layered pups is certainly one which I would like to hear more about. (and also I hope that it may one day be possible to have that structure grafted into some sort of ROM media for boot speed).
I don't have the big RAM that you are used to but as time moves on it is going to be commonplace to have sufficient RAM to squeeze a very fat pup and a whole heap of sfs addons (including a 'personalised' sfs) into fast memory so I'm keen to see techniques for this expanded and discussed.
As far as I can tell the Puppy layering setup lends itself to adding a personalised sfs over the top of the main sfs, although that is not the traditional puppy way - normally it is the personalised rw savefile that populates the top layer, but I am keen to see other ways of running that personalised layer as an sfs (readonly) for the sake of system stability.
Don't know if you ever try to run your pups that way but it is what I do with my daily driver pup and the concept of readonly layered pups is certainly one which I would like to hear more about. (and also I hope that it may one day be possible to have that structure grafted into some sort of ROM media for boot speed).
I don't have the big RAM that you are used to but as time moves on it is going to be commonplace to have sufficient RAM to squeeze a very fat pup and a whole heap of sfs addons (including a 'personalised' sfs) into fast memory so I'm keen to see techniques for this expanded and discussed.
-
stemsee
I enjoyed the talk, of course the dynamics in the live room cannot be conveyed in a statically shot video.
Nerves are good for performance, but tend to cause a rushed performance. Slow down a tad.
If the audience doesn't chuckle at your humorous slides, ask them "Why isn't that funny? I though it was funny." dwell on it for a moment ... such interactions will make the presentation more memorable.
Did you mention methods of installing packages and package format? SFS, yes, .pet, .deb, .rpm, .txz, .tgz ? Did you mention compiling and creating new2dir -> pet package (puppy native file format)? I did skip some places so maybe I skipped that.
sfs compression algorithms, gz, xz, lz4. Did you mention Remastering the running system for completely personalised distribution? Also heavy reliance on bash scripts.
Nerves are good for performance, but tend to cause a rushed performance. Slow down a tad.
If the audience doesn't chuckle at your humorous slides, ask them "Why isn't that funny? I though it was funny." dwell on it for a moment ... such interactions will make the presentation more memorable.
Did you mention methods of installing packages and package format? SFS, yes, .pet, .deb, .rpm, .txz, .tgz ? Did you mention compiling and creating new2dir -> pet package (puppy native file format)? I did skip some places so maybe I skipped that.
sfs compression algorithms, gz, xz, lz4. Did you mention Remastering the running system for completely personalised distribution? Also heavy reliance on bash scripts.