racket

General talk about packaging procedures and packages.
User avatar
mimosa
Salix Warrior
Posts: 3101
Joined: 25. May 2010, 17:02
Contact:

Re: racket

Post by mimosa » 4. Oct 2012, 11:44

So what could it be? It builds on laprjns's machine, just not mine. And there is no possibility of user error that I can see.

User avatar
mimosa
Salix Warrior
Posts: 3101
Joined: 25. May 2010, 17:02
Contact:

Re: racket

Post by mimosa » 4. Oct 2012, 12:07

(if memory serves)
... no, it doesn't. However, the errors in the log file relate to that part of the script. It's a mystery ...

Interestingly (though I don't know what to conclude from it) my build with the relative path for the symlink (exactly as you said) has incorrect paths for the help files, just like the indirect path for $bindir, pointing to the location in the build directory.

FWIW:

Code: Select all

root[laprjns]# cat *log
tar -xf racket-5.3-src-unix.tgz
./build-racket.sh: line 173: $'\302\240': command not found
./build-racket.sh: line 174: $'\302\240': command not found
./build-racket.sh: line 175: $'\302\240': command not found
./build-racket.sh: line 176: $'\302\240': command not found
./build-racket.sh: line 177: $'\302\240': command not found
./build-racket.sh: line 183: $'\302\240': command not found
build() failed.

User avatar
laprjns
Salix Warrior
Posts: 1039
Joined: 28. Aug 2009, 01:30
Location: Connecticut USA

Re: racket

Post by laprjns » 4. Oct 2012, 16:20

mimosa wrote:The only thing I can think of is maybe racket doesn't like being built for i486 with a 64-bit processor. To test that, I will try and find room on my wife's machine for a temporary beta install. I tried asking the people at Racket, but no luck there so far.
For building and testing Slaix 14 packages I multiboot both 32 and 64 bit version on a 64 bit quad core machine

I
mimosa wrote:nterestingly (though I don't know what to conclude from it) my build with the relative path for the symlink (exactly as you said) has incorrect paths for the help files, just like the indirect path for $bindir, pointing to the location in the build directory.
The symlink has nothing to do with your path problem. I believe that I mention that the paths to the help files had the same path problem when I first posted about the problem back a few days ago.

Are you having similar problem on other apps that you are building on your 32-bit machine?
“Don’t you see that the whole aim of Newspeak is to narrow the range of thought?"

User avatar
laprjns
Salix Warrior
Posts: 1039
Joined: 28. Aug 2009, 01:30
Location: Connecticut USA

Re: racket

Post by laprjns » 4. Oct 2012, 16:37

mimosa wrote:Meanwhile, I have reinstalled the beta with a few home comforts, and I'll test the package with the relative path to the symlink target
I'm not sure what you mean by "a few home conforts", but you may tried skipping this and just do a test build with a fresh install and just add the minimum packages needed to build the package. I don't think there is any build dependancies for racket so a Basic install may be all you need.
“Don’t you see that the whole aim of Newspeak is to narrow the range of thought?"

User avatar
mimosa
Salix Warrior
Posts: 3101
Joined: 25. May 2010, 17:02
Contact:

Re: racket

Post by mimosa » 4. Oct 2012, 16:39

The symlink has nothing to do with your path problem.
Quite so. What I mean is, I got the symlink *right* (and it works for you), yet still I get incorrect paths - and they look uncannily similar to the wrong $bindir path, that is, they are absolute paths to wherever that stuff resides in the build directory, not relative paths to wherever it should get installed to.

No no problems as far as I am aware, whether building for either architecture on this 64-bit machine, or earlier for 32 bits only using a 32-bit motherboard. Apart that is from various problems of my own making, which were eventually solved ...

This racket is a much more complex animal though than anything else I've built. I know for instance that it creates a core version of itself to build the one that will be installed. As a result, it uses scripts in scheme/racket to do so - one of these is referred to in the drracket launcher script. Consequently, the way paths are generated is not transparent.

Gapan says there's no point trying it on another machine (though in a way, I've "tried" it on yours ;) , and it worked). If that is so, the logical conclusion is I'm doing something wrong - but I've gone to some lengths to exclude any such error. In particular, the build environment doesn't come any cleaner than on my last attempt.

I'm stumped.

User avatar
mimosa
Salix Warrior
Posts: 3101
Joined: 25. May 2010, 17:02
Contact:

Re: racket

Post by mimosa » 4. Oct 2012, 16:40

a few home comforts
No, last time, I did a core install, changed nothing, built it three ways overnight.

User avatar
mimosa
Salix Warrior
Posts: 3101
Joined: 25. May 2010, 17:02
Contact:

Re: racket

Post by mimosa » 5. Oct 2012, 11:39

I have got it to work, but I'm not really sure how, because I have discovered that although /usr/bin/drracket is created with a $bindir path to somewhere in the build directory *during the build*, if all goes well, it is changed to the right path at the end. However, that wasn't happening originally. Because it's quite a long build on this hardware, I was saving time by examining the file during the build to check the path. So any number of previous builds I chalked up as failures may not have been in fact - though that doesn't apply to the three I left running in parallel overnight the day before yesterday, my most recent prize failures. Some change, somewhere along the line, fixed it.

My one definite success passes

Code: Select all

--docdir=/usr/doc/$pkname-$pkgver
to configure, in an attempt to avoid doing that manually. It didn't have any effect on the location of the doc files, but the build was otherwise successful.

User avatar
mimosa
Salix Warrior
Posts: 3101
Joined: 25. May 2010, 17:02
Contact:

Re: racket

Post by mimosa » 5. Oct 2012, 21:39

I have posted an apparently successful i486 build at my Salix repo page; I'll set it building for the other arch overnight, but given that worked from the outset, I don't expect any problems.

If anyone who has been helping me with this little project feels inclined to give it a whirl before I submit it at Sourceforge, please feel free.

I tried with both absolute and relative paths for the symlink target, but only the former worked. This way everything, including the files normally covered by the doc line in the SLKBUILD, is in the right place (though because of the symlink, it's quite hard to tell), and the help is informative about where to find the licence.

Phew.

User avatar
mimosa
Salix Warrior
Posts: 3101
Joined: 25. May 2010, 17:02
Contact:

Re: racket

Post by mimosa » 23. Jan 2014, 23:23

I'm no longer using Racket myself, and given the difficulty of building the package last time round, as well as the fact that it seems to have lost its way a bit as a teaching project, I'm inclined to drop it for 14.1. However, are there any users out there?

Post Reply