woof-CE needs you

[Iguleder]

I am not talking about a 300 page manual.

I see your point - yes, if woof-CE fails to produce a working Puppy, it's a problem. However, we can't just document all possible problems with woof-CE as a series of questions and answers, because there are so many things that can get screwed up. It's a matter problem solving skills and attitude.

[666philb]

Hello all.

If it may speed development of the Jessie part of woof -- or just if you
somehow need these libs --, please find attached the list of libs found in
saintless & Co.'s DebianDog-Jessie.

I don't know if it's ok to do it like that, but it's saving a lot of time.

I've transfered them to the interim puduan pupsave file. For your
convience, here is a download URL for this pupsave containing the libs
found in saintless' DDog-Jessie.
-- https://www.adrive.com/public/tZYvuk/pu ... rf.2fs.lzo --

Now to hunt down the other half!

BFN.

musher0

hi musher0,

the main goal is to have a DISTRO_PKG_SPECS that will produce a fully working pup with nothing missing and of a release quality.
this doesn't happen all at once! but is something to gradually progress towards.

once you've added some missing stuff to DISTRO_PKG_SPECS i would use the woofgui and then run 'download databases' (this is 0setup) and then 'download packages'(this is 1download). this will be a lot quicker than the first run as most things are already downloaded.

then on the 'create packages' tab you can choose to build the individual missing packages withtout having to create them all again.

some things to note:

when running download databases/0setup if you have already recently downloaded the databases you can skip downloading them again but still run this script as it sort of resets woof, and gives you info quickly if you've named anything wrong in DISTRO_PKG_SPECS.

after running download packages/1download ... check to see if there's a 'downloads failed' file and read it.

after running create packages/2create packages .. check to see if there's a 'error creating packages' file. (it's called something like this)

once you've started making changes to the DISTRO_PKG_SPECS be sure to make a copy of it before making anymore changes. one misplaced character in it can make woof fail and be a pig to find.

use the PPM to find the correct names of packages that need to be added to DISTRO_PKG_SPECS

----------------
18 months ago i was in exactly the same position as you

[mavrothal]

once you've started making changes to the DISTRO_PKG_SPECS be sure to make a copy of it before making anymore changes. one misplaced character in it can make woof fail and be a pig to find.

It is actually a good idea instead of making your changes in the woof-out folder to make them in you local git tree, commit them and run merge2out again.
So a) you can see if anything was wrong with your changes easily by checking only the commit changes and b) revert easily if totally messed up. Downloaded packages are not affected by the process but your semi-ready puppy root folder will be deleted. So may want to take out of the old woof-out, run merge2out and reintroduce it.
When happy with the outcome you can easily contribute your changes back to woof

[musher0]

Thanks, phil and mavrothal,
for the advice and the encouragement.
I'll study your posts more closely then.
BFN.

[Ted Dog]

still wanting advice on which WolfCE based distro to use as a build base... If you all are embarrassed about tooting your own distro PM me! I will even consider going to the dark side if needed for this first attempt.

[slavvo67]

I don't mind the creation process not working perfectly but it gets disappointing when the first try doesn't go as planned.

Perhaps a walk-through with one Puppy Distro that is known to work near flawlessly once created?

Then, I know that I'm doing it right the first time and any others I attempt to create would more thank likely be items to fix rather than user (my) issues.

Very much appreciated,

Slavvo67

[01micko]

still wanting advice on which WolfCE based distro to use as a build base... If you all are embarrassed about tooting your own distro PM me! I will even consider going to the dark side if needed for this first attempt.

Be patient my man. I'm doing a brand new build of slacko64. I'm making sure it can be fully reproducible. That would be a good jumping off point if you want to leverage Barry's T2 64 packages.

[wanderer]

in my humble opinion

the woof-ce gurus should build a minimal woof-ce version
with just the basics
jwm rox dillo gtkedit etc
this will allow download times and complexity to be minimized
and can be tweaked until it works reliably
this will help non-gurus to learn the system
and learn how to add to it

wanderer

[smokey01]

I tried the woof-CE build system some time back but had a few problems so I gave up.
As there seems to have been some changes I thought I would give it another crack.

Using the git command in Fatdog64-700 didn't work for me so I tried Fatdog64-701 and it didn't work there either.

Code: Select all

git clone https://github.com/puppylinux-woof-CE/woof-CE.git -b testing

Anyway I tried slacko-5.7 it it worked.

I then cd to woof-CE and ran ./merge2out. That seemed to work. I then cd to woof-out_x86_x86_slackware_14.1 and ran ./0setup. So far so good.
Now I ran ./1download, it took some time, probably about an hour. After that I ran ./2createpackages which took 1.5 to 2 hours. At the end of this step it reported errors were logged in a file called ERROR-2CREATEPACKAGES so I had a peek.

This is the contents of the file:

ERROR: packages-pet/dropbear-0.52-i486.pet does not exist.
(Generic name: dropbear)
You will need to find this PET package and place in packages-pet.
Do it, then rerun this script and choose to build dropbear.

ERROR: packages-pet/dropbear_DOC-0.52-i486.pet does not exist.
(Generic name: dropbear)
You will need to find this PET package and place in packages-pet.
Do it, then rerun this script and choose to build dropbear.

ERROR: packages-pet/gmeasures-0.7.pet does not exist.
(Generic name: gmeasures)
You will need to find this PET package and place in packages-pet.
Do it, then rerun this script and choose to build gmeasures.

ERROR: 'guess_fs' package does not exist.
You will need to find a matching package and place in packages-pet,
or packages-tgz_txz-14.1 as appropriate.
Do it, then rerun this script and choose to build guess_fs.

ERROR: 'pburn' package does not exist.
You will need to find a matching package and place in packages-pet,
or packages-tgz_txz-14.1 as appropriate.
Do it, then rerun this script and choose to build pburn.

ERROR: 'pmusic' package does not exist.
You will need to find a matching package and place in packages-pet,
or packages-tgz_txz-14.1 as appropriate.
Do it, then rerun this script and choose to build pmusic.

ERROR: packages-pet/sfs_manager-0.8.pet does not exist.
(Generic name: sfs_manager)
You will need to find this PET package and place in packages-pet.
Do it, then rerun this script and choose to build sfs_manager.

ERROR: packages-pet/wv_DOC-1.2.5-w52.pet does not exist.
(Generic name: wv)
You will need to find this PET package and place in packages-pet.
Do it, then rerun this script and choose to build wv.

ERROR: packages-pet/x264_DEV-2.012-i686.pet does not exist.
(Generic name: x264)
You will need to find this PET package and place in packages-pet.
Do it, then rerun this script and choose to build x264.

ERROR: packages-pet/z_fixcache-0.1ff-noarch.pet does not exist.
(Generic name: z_fixcache)
You will need to find this PET package and place in packages-pet.
Do it, then rerun this script and choose to build z_fixcache.

I think I managed to find one of the missing pet packages and rerun the script but before I realised I could have just typed in a single package the script ran the entire process again for another hour and CTRL-C would not stop the script so there goes another hour of my life I won't see again.

When the ./2createpackages script had finished it gave me the same error list minus one package, oops two packages. I actually renamed sfs_manager-0.7.pet to sfs_manager-0.8.pet to see if would comply and it did. The problem is there may have been some important changes in the version upgrade.

Ah the other pet I found was z_fixcache-0.1ff-noarch.pet.

After scratching my head for a few minutes I said to my self, self, these missing packages look a little familiar. I am pretty sure dropbear, pburn and pmusic were all missing last time I tried this.

I managed to find all of the missing files albeit different versions.

So my questions are:

1. What files do I need to edit to make it recognize the later pets, and
2. How come it's required if the system is being maintained or is this what Iguleder means when he says 'woof-CE needs your help too'

Cheers

[smokey01]

Not so long ago Linux was not all that popular and it wasn't because it was not as good as it's competitors, it just seemed a lot scarier or unfriendlier to use, or that was the general consensus anyway. We who have been using it for years know different. Often when I mention Linux to people they still think it's driven from the command line which turns people away.

I think these builds systems are a bit similar. Many people take a look and say, wow, that's way above my knowledge level. The fact is writing the system probably is but driving it shouldn't be. You don't have to be a mechanic to drive a car but it helps.

In the past I tried the porteus build system and found it easy and very user friendly. If we want people to be involved then they have to be comfortable.

Before anyone says there is already a GUI for woof, I know but it's busy and still confusing for newcomers.

My vision is a single page gtkdialog GUI that lets you choose your system using checkboxes and comboboxes. There would always be a long list of mandatory programs that need to be included where the user would have no discretion. All of the alternative apps could be user selected. Another benefit would be not having to download all of the apps, just the ones to be included.

Like most things, this is a lot easier to say than code. Although woof and woof-CE are great tools they are way too scary for the average user. I acknowledge the developers have done a fantastic job and understand they would like to see more people using it but I just don't think it's going to happen in it's current form.

Now before anyone gets all bitter and twisted this is simply my opinion.
As the saying goes, my 2c worth.

Ah, and yes I am willing to be involved. My comments are intended to be constructive not critical.

Regards

[Ted Dog]

Ok thanks will look forward to Slacko64 to practice building a slack version.. then press on to T2

[01micko]

Hey Ted Dog

Must stretch that patience a little further. Power outage for me today.

[smokey01]

I'm interested in giving slacko64 a go.

[Ted Dog]

Hey Ted Dog

Must stretch that patience a little further. Power outage for me today.

UPS... and not the United Postal Service one!

Have one on my main computer that is not a laptop. Your stuff is too important to lose, wonder if AmazonPrime offers free shipping to QLD. (Aussie right???)

[01micko]

Power is back up.

Was funny, they lobbed on the job at 0830, sat down for smoko from 0900 til 0930 (plus a little). Took my dog to the vet (got a new one, poor old Bill - the one imortalised in my avatar passed on a year ago). Got back from vet at 1130, power restored at 1200 - luckily before lunch! It would have been 1300 (+) otherwise!

Dunno about amazon - my son has an acc, I'll ask him.

Smokey and Ted Dog, I'm trying to tune it up now - looking good so far, so when I say the word - go for it! (lots to test though and as always there will be stuff to recompile).

[mavrothal]

I tried the woof-CE build system some time back but had a few problems so I gave up.

<snip>

So my questions are:

1. What files do I need to edit to make it recognize the later pets, and
2. How come it's required if the system is being maintained or is this what Iguleder means when he says 'woof-CE needs your help too'

The problem is that puppy builders are more eager to get a nice puppy out (good thing I believe) than maintain a reproducible puppy building system.
Missing packages for example. A developer updates a pet but never puts it in the ibiblio repo (and some times never puts it anywhere!). Then you are simply stuck. Same is true for "little changes" in woof scripts that never make it back to woof.
So how can someone help? By bringing up exactly these problems!
The seasoned developer can easily work his/her way around them and often does not bother to fix them properly.
For example more than a year now Mick knows that some slacko pets are not in the repo, but did not have the chance to upload them there. I'm sure that if a couple more people scream about it, he will or edit the recipe if outdated. (Sorry Mick for giving you up )
Similar issues will happen with Tahr (bug Phil ) or Libre (bug Iguleder )

On the other hand would be nice if the aspiring woofer has some knowledge in scripting and debugging and is comfortable with the terminal so (s)he provides specific and detailed feedback and hopefully solutions.
A new woofer after building an already build puppy will start changing things to his/her liking and things will likely break. Would be nice to have as much as possible specific information then (including a patch for the changes ) so can figure out what might be the problem.
Woof builders fulfilling the above may also want to directly address possible issues in the woof-CE issues page (or at least check that they do not rediscover the wheel)

[smokey01]

Thanks mavrothal.

I did find most of the missing pets in the Internode repository which of course is mirrored from ibiblio. The problem, I think, is the woof scripts need updating to recognise the newer packages.

How do I or someone else do this?

BTW it would be dangerous if every Tom, Dick or Harry had the ability to change things in woof, it would soon be destroyed.

Cheers

[mavrothal]

The problem, I think, is the woof scripts need updating to recognise the newer packages.

How do I or someone else do this?

Have to change the corresponding DISTRO_* in wiif-distro/<ARCH>/<DISTRO>/<VERSION>

BTW it would be dangerous if every Tom, Dick or Harry had the ability to change things in woof, it would soon be destroyed.

Below are people allowed to change things in woof. AFAIK non of them is named Tom, Dick or Harry .
However, more people would be welcome in the team. It mainly depends on previous contributions and approval from existing members.

I should also point out here that sometimes people get discouraged or even annoyed if a proposed change is not accepted and give up. There are several reasons why this may happen. For example a change may be good for the running puppy but affect negatively the build system. Or may just be a work around instead of properly addressing the issue (and discussion goes on till a proper solution is found). Or maybe they are unsure how things will be affected and ask for some real life tests (ie a pet to be tested in a current puppy). Or may be simply a matter of policy and goals (someone may suggest to built only for 64bit systems for example).
Woof changes is not a "take it or leave it" approach. Actually, not only every pull request can be discussed but every single line of code can be discussed in github. For this reason, it is not uncommon that even woof members that can commit their changes directly to woof without discussion, issue a pull request so a change that is not a bug fix or something simple, can be exactly discussed.
I should point out here that you do not need to be woof member to discuss an issue. You just need a github account.

[Iguleder]

Just in case this isn't clear enough: anyone can contribute to woof-CE by forking it and sending a pull request.

Members of woof-CE can commit directly to woof-CE, without having to create pull requests first. However, most major changes that may affect multiple Puppy flavors are either discussed first (through the forums, a GitHub issue, a comment on a commit ...) or merged using a pull request for peer review (i.e. if you want someone to take a look at your changes before they are merged). Trivial changes with zero risk of breakage (like updating fonts, removing old wallpapers nobody uses anymore or small bug fixes) are merged directly, since it's quicker.

Contributors that are serious about contributing to woof-CE can always join the woof-CE GitHub organization, to gain that direct commit access. The risk of damaging woof-CE is minimal, because changes can be reverted and users can be put in the "members" group, so they cannot do wild things like deleting the entire woof-CE. I'm a "member", too

woof-CE is a great platform for team work. It allows us develop Puppy releases as a community and discuss things in a non-linear form (in contrast to a forum thread). Here's a very recent example - 01micko released a new version of netmon_wce and notified me via a comment on the commit where I added it to Librepup. Now I know that I need to build a new PET and upload it to the repo, for the next release of Librepup. In the forums, this kind of collaboration is pretty much impossible: many puplets have outdated, in-house Puppy packages (things like pMusic, the tray icons, etc') because of this.

I believe the upcoming wave of Puppy releases will be super awesome, thanks to all improvements introduced in woof-CE recently. It's amazing to see how collaboration of several developers on infrastructure affects the end result.

[mavrothal]

anyone can contribute to woof-CE by forking it and sending a pull request.

Absolutely!

Here is a possible workflow for this:
- Get a github account (A free one - No credit card necessary)
- Fork puppylinux-woof-CE/woof-CE to your account (will become <user_name>/woof-CE)
- Clone YOUR fork to your computer (need to have devx loaded). Make sure you have configured git in your machine to your liking.
- Build your puppy and/or just make any necessary changes in your local woof-CE git folder (not only your woof-out folder)*.
- Commit changes in your local git. (One commit per change. NOT one huge commit with everything in it)
- Push changes from your local git to YOUR github fork
- Issue a pull request from your fork to puppylinux-woof-CE/woof-CE
- Done!

Besides the above links, the web (and the forum) is full of HOWTOs for each of those steps. Just look around a bit.

* If you already have made changes in your running puppy that you would like to contribute in woof-CE you can also use the patch generator and then patch your local git with them to add the changes in your woof-CE copy.