[Coco] nitros9 daily clean build script for Gene

Tormod Volden lists.tormod at gmail.com
Thu Oct 23 16:01:17 EDT 2014


On Thu, Oct 23, 2014 at 2:42 AM, Gene Heskett wrote:
> On Wednesday 22 October 2014 19:30:19 Tormod Volden did opine
> And Gene did reply:
>> On Thu, Oct 23, 2014 at 1:06 AM, Gene Heskett wrote:
>> > On Wednesday 22 October 2014 16:37:09 Tormod Volden did opine
>>
>> > make it anyway gets me this:
>> > gene at coyote:/opt/nitros9-2014-10-22-18-46$ sudo make dsk dskcopy
>>
>> Nobody told you to use "sudo". This is Linux beginner fail number one,
>> padding on "sudo" whenever there is an issue. If you need sudo to
>> build nitros9 then your system is severely messed up. That's like
>> using an administrator account on Windows to read e-mail. Hey wait...
>> Just kidding :)
>
> I don't have creation rights in /opt, and that is where the script runs.
> But the script chowns it back to me:www-data after each download/build.
> That way apache2 can also access it if you want to take a look with your
> browser.  Link in sig of course.

There are many ways to sort this out properly, without executing
unsecured scripts from an internet server with full administrative
rights on your computer.

sudo chown $USER /opt

is a better alternative, if you really need to do this a system folder
like /opt.
If your webserver needs it to be owned by root, do a
sudo chown root /opt
after running the "daily" script.

Note there is no -R option to chown there! As long as the files are
readable by "other", apache can read them even if they belong to you.

>
>> > **************************************************
>> > *                                                *
>> > *              THE NITROS-9 PROJECT              *
>> > *                                                *
>> > **************************************************
>> > make -C /lib && make -C /level1 && make -C /level2 && make -C /level3
>> > && make -C /3rdparty && :
>> > make[1]: Entering directory `/lib'
>> > make[1]: *** No targets specified and no makefile found.  Stop.
>> > make[1]: Leaving directory `/lib'
>> > make: *** [all] Error 2
>>
>> This is because you have set NITROS9DIR to an empty value.
> No, $NITROS9DIR is set in my bashrc, and that meant it was not pointing at
> my clone+date, but to /opt/nitros9.  That has been a part of my .bashrc
> for a good decade.
>> You
>> shouldn't set it all. Use "unset NITROS9DIR" in your current shell, or
>> start a new fresh shell.
>
> That would inherit the assignment from my .bashrc.
>>
>> > I am going to comment the lwtools branch out of my script, which
>> > makes it much like yours, then try yours again, and see if I have
>> > any makefiles.
>>
>> Please use my script, or follow the instructions. It is impossible to
>> help you if you insist on doing things your own way.
>
> Are you saying that a fresh hg clone is not the same as a later
> hg pull;hg update?  I can do an hg pull and get nothing afterwards.  If I

hg pull; hg update will give you the same as hg clone. Yes! That's
exactly why I recommend you to do "hg pull; hg update" instead of
cloning the whole repo from scratch all the time! Saves time and
bandwidth and server and tropical forests.

As long as you keep this repo pristine, and only build and develop in
copies from it, it is guaranteed to stay clean and consistent and it
will update cleanly every time.

> understand hg, and I'm about a first grader at it, I have assumed the
> clone to a fresh directory should give me the cleanest copy, and indeed it
> did until it ran into the blanked makefile in the level3 directory, the
> one you now have a replacement for in your inbox.
>
>> > This is frustrating at best.
>>
>> Do not complicate things with build scripts if you cannot get it to
>> build by hand. The whole building from scratch is fairly simple. See
>> http://sourceforge.net/p/nitros9/wiki/Building_NitrOS9/#obtain-and-buil
>> d-nitros-9
>
> Tormod, buildit scripts are my escape from the short term memory problems
> of being 80 years old.  When you are my age, you will depend on them
> simply because they do the same thing every time even if you have to read
> the script to refresh your own memory just to make sure you are about to
> fire off what might be the wrong one.

I have more around half of that age, but I don't remember this stuff
either! But I look up the documentation, even the one I wrote myself
some months ago. Memory is not the main problem. But age often brings
stubbornness...

>
>> Once you have this under control, you can think about making your own
>> changes, using build scripts, and so on. One step at at time.
>
> and all those 1 step at a time things were resolved and committed to a
> script in about 2001 maybe.  The script of course gets updated when I get

You expect your old build scripts to work for years and years. They
won't. Things change, so you must check the documentation like
everybody else.

> the message its no longer correct.  I don't recall being told that the
> temporary lwtools branch was now frozen, and that work was now in the main
> tree.

That happened in January and I posted a message called "nitros9
development now in *default" branch of repo". And I always said it
would be done before releasing 3.3.0. And I mentioned it in my reply
to you from that time, in the thread I linked to yesterday :p

Never mind, now we are all on the same page (branch) and I am looking
forward to your patches.

Cheers,
Tormod


More information about the Coco mailing list