Nekochan Net

I got my Origin running at my new place, got my internet up finaly, so I wanted to get back at porting. Woohah!!! :-D
BUT! I downloaded the current beta packages and I looked into them... and there is absolutely no standard being used. I wanted to do something like 'install *.opt.relnotes' and the likes to install the porting docs and patches 'n stuff. But everybody uses different names and different packaging styles. This should realy be resolved quickly before we have like 79854 packages all made in different ways. And then it'll be one hell of a job to fix that.

And the apps 'make' and 'render' install as user '60000' with group 'user' instead of 'root/sys' ?

I see the need for some serious whipping.

_________________
Shall I describe it to you? Or do you want me to get you a box?

Actualy,

I see the need for a central build server thing where one person would receive the ported programs sources/patches including build instructions after the porter finished his job, and then he'd build + package them properly and then put them on the site, after testing that it worked ok.

As a matter of fact I'd gladly take that responsibility. Because the way it looks now... I'm not going to install all those differently behaving packages just to do everything all over again after they're changed and changed again. Things need to be well organised. There are a load of people here who are great at porting, and I'd be happy to regulate all those outputs into nice standard canals. My Origin is eager to do those massive test/build/package runs for you all :-D

_________________
Shall I describe it to you? Or do you want me to get you a box?

Whiter to the rescue! That's a royally cool proposition.

_________________
覇気元
Eroteme.org

I thought a standard had been agreed on, awh well no doubt we'll have several more pages of dicussions on where stuff shall go
I think a "porting rules document" would be a good idea. That way everyone could work off the same hymne sheet

_________________
"Ever tried. Ever failed. No matter. Try Again. Fail again. Fail better"

yeah, such a 'general cheat sheet' would be very nice!!

_________________
r-a-c.de

Awh well ye guys actually do the work (well i'm considering getting a mipspro license) so ye might as well start here.

What exactly do we need in such a document.

Compiler options?
PATH?
MANPATH?
etc.

_________________
"Ever tried. Ever failed. No matter. Try Again. Fail again. Fail better"

I see some serious stumbling blocks here.

I can't speak for everyone here, but my tardists are patterned after the SGI Freeware builds. Why? A few reasons:

1. Descriptive package names. These descriptive names carry through to all subpackages. When I do a 'versions long' I want to be able to see in detail EXACTLY what is installed on my system, no the vague "Software", "Man Pages", etc. that the automated packager currently provides. This is essential to me.

2. Postop/Preop scripts. Incredibly important, and I take full advantage of these to set up install and postinstall environments, generate host keys if necessary in the case of openssl, etc.

3. Config file protection - It is absolutely essential that config files are protected and only updated if the user hasn't made any changes to them. This is important for packages such as bind which could potentially clobber a user's entire domain after installation, or turn the chkconfig setting back off.

These are incredibly important things to consider for overall quality. Creating a package isn't just about how quickly you can slap it together and how convenient it is for you as a developer to install porting notes. Quality is key. Think of the user.

I'm not against going to an automated layout, but I feel it would be too easy to drop the ball at this point. I prefer to hand tune my packages so they meet my needs stated above and hopefully give the end user a little respect as well.

We already have individuals on comp.sys.sgi stating in public 'avoid Nekoware', let's not make it worse.

_________________
Twitter: @neko_no_ko
IRIX Release 4.0.5 IP12 Version 06151813 System V
Copyright 1987-1992 Silicon Graphics, Inc.
All Rights Reserved.

avoid nekoware? what sorta problems do they have with it?

_________________
"Ever tried. Ever failed. No matter. Try Again. Fail again. Fail better"

Apparently it's not SGI Freeware and hence not to be trusted.

_________________
Twitter: @neko_no_ko
IRIX Release 4.0.5 IP12 Version 06151813 System V
Copyright 1987-1992 Silicon Graphics, Inc.
All Rights Reserved.

no, let's be honest about this ... in some cases there are problems. Doing something people can depend on demands discipline. Unsupervised group projects are not renowned for discipline. At this time there is no way to even know pre-installation whether the app will run or which level of Irix is required or or or. I don't mind futzing around with my computer but some people are not so ... adventurous ? There is a big difference between kewl d00dz hacking up a storm and userland-ready applications.

If you want to do the one, spiffy. if you want to do the other, equally spiffy. Just don't confuse the two.

We've had this argument before and it's the primary reason I've pretty much stopped trying to create new tardists. Yes you've inspired me to pretty much quit. I can't be blamed for changes that SGI has made to IRIX that makes compiles under 6.5.22 not work on older versions and I also can't be expected to downgrade systems I use on a regular basis just for the hell of it.

The reference in question wasn't about compatibility anyway, it was about lack of trust - as in malware - openssl/openssh with backdoors and other tinfoil.

Hmm, yes on second thought sure, make Nekoware anyway you like whiter - I've grown weary of all the arguing.

_________________
Twitter: @neko_no_ko
IRIX Release 4.0.5 IP12 Version 06151813 System V
Copyright 1987-1992 Silicon Graphics, Inc.
All Rights Reserved.

So? I mean, yes, maybe there are some problems, but isn't it worth trying to make it to work? What's the other alternative? Nothing... sgi Freeware are from 8/03, so anything more recent is definitely welcome, even if it might require some work to make to work and sometimes it might not work at all...

I for myself do appreciate you guys' hard work... If it works, then great. If it doesn't, I won't hold any grief against anybody... Software development requires tremendous discipline that's right, but I think the community is expecting too much from you guys! I have low expectations and am usually happy & surprised when it works right away... Not because you guys can't make it to work, but just because you can't have on of each hardware/OS combination to test with...

It not like it's expensive commercial applications anyway...

Do you consider every comment which is not praying before the altar do be an argument ?



There are what, 500 people in here now ? How many have extensive Irix experience ? How should they feel when they download the spiffiest new xyz only to find that it doesn't work ? All I said before about that, was that things for general consumption SHOULD have a decent readme. If xyz won't run on irix 6.5.20 (the latest available one) then it would seem that a readme TELLING people that ahead of time would be a good idea. If that is your idea of an argument then I don't know what to say ...



Can you find any comments to the effect that this is *your* fault ? Not from me ....



Nor will you find any sugggestions like that from me ... but if Nekoware is to be of any use to the general community, then it should at least tell people what's required before they go to all the trouble of trying to run it. There's nothing wrong with being a hot-dog hacker dood .... but if nekoware was intended to be good for the entire group of people running these things, then a little documentation would seem to be a requirement. And maybe the newest widely available Irix version would be good, too. You're right, I did comment on that. And you guys said that wasn't important to you so I shut up about it. End of argument from my point of view. You haven't seen a single remark about that decision from me since. In fact lemme set myself to readonly, since any different opinions seem to upset you.





well, if one had a *serious* need for security they'd probably be correct to suspect anything that wasn't compiled under their own control. That's not a cut at the people here, just a general observation about security. Like sausage, if you need to really trust something, better make it yourself.



Excuse me for having any ideas. Sorry.. Will go read-only and see if I can sign up for some kou-tou lessons. Perhaps that will stop all the arguing. See ? Easy solution leaving everybody happy. Life just isn't that hard, ya know.

hello,

expected me to answer now?

i can understand neko and why he stopped releasing new tardists.
but this is how it works. some people have problems and get nasty not seeing that all the work
is just for fun and voluntarily. some others and myself spend hours to provide some nice and free
soft just to support less versed users but sometimes some people behave like they would have
paid many bucks for the app incl. support.
this sucks but is also a part of the whole thing.

i've worked for a software company for a few years and it was exactly the same.

but there're also other examples. many users/members are grateful and happy. and sometimes
one can have nice and productive contact with those people using the apps.(hello hamei ).

however, there're always 2 sides. but for me it's worth the trouble as long as i can cause some people to be a bit more happy with their machine.

i thought of including a readme with each package. no matter which format. i guess this is much
more flexible like fixed deps from within the tardist.

btw: you can't compare nekoware and the other stuff with sgi's freeware. look how fast we are.
how many new versions do we provide each month?

live long and prosper

_________________
r-a-c.de

woah. amazing how bits'n'bytes can shift the emotions so much.
Eheh :-)

What I meant to say:

if you look at any serious bsd or linux distribution, there are a lot of people porting software for it. Which is GREAT and realy needed.
But those people don't make the packages that are put on the download servers. They submit their stuff to a server where someone else will do the final build/package thing, to make sure everything will work/look the same.

I thought of this as a nice idea to implement here. But if not, then ohwell, we'll just do all what we do best: port software and build tardists. :-)

On another note: people in the newsgroups stating 'avoid nekoware' ?
Damn. Those guys are just weird. Let's ignore them :-)

_________________
Shall I describe it to you? Or do you want me to get you a box?