On Thursday, Feb 20, 2003, at 22:35 US/Eastern, Ben Hines wrote:
Apple's curl is more than two years old (!) Fink never even had such
an old perl, we started at 7.8.1 in Aug 2001.
That may be true. If Fink relies upon behavior in the "modern" curl--
my later message indicates there is still so
On Thursday, Feb 20, 2003, at 14:29 US/Eastern, Benjamin Reed wrote:
On Thursday, February 20, 2003, at 02:23 PM, Bill Bumgarner wrote:
curl -f -L -s -S -P - -O
ftp://ftp.trolltech.com/qt/source/qt-x11-free-3.1.1.tar.bz2
Trolltech's server is anal and only allows passive connections. You
curl -f -L -s -S -P - -O
ftp://ftp.trolltech.com/qt/source/qt-x11-free-3.1.1.tar.bz2
More curl problems -- same behavior with /usr/bin/curl or /sw/bin/curl.
I'm beginning to dislike curl.
[bumbox:/tmp] bbum% curl -v -f -L -s -S -P - -O
ftp://ftp.trolltech.com/qt/source/qt-x11-free-3.1.1.tar.
On Thursday, Feb 20, 2003, at 12:14 US/Eastern, Alexander Hansen wrote:
But you do have curl-ssl installed, which will give you /sw/bin/curl .
Oh, duh! Yes.
---
This SF.net email is sponsored by: SlickEdit Inc. Develop an edge.
The most com
On Thursday, Feb 20, 2003, at 11:37 US/Eastern, Martin Costabel wrote:
On jeudi, fév 20, 2003, at 14:58 Europe/Paris, Bill Bumgarner wrote:
[]
Right. Your words make perfect sense, the situation does not.
Have you tried this with /usr/bin/curl instead of Fink's curl? Does it
behave i
On Thursday, Feb 20, 2003, at 02:01 US/Eastern, Martin Costabel wrote:
Bill Bumgarner wrote:
[]
I would think that answering 'n' -- disabling passive -- would work
much better with firewalls. It is certainly a requirement for NAT
translating routers (airport base stations and
On Wednesday, Feb 19, 2003, at 20:38 US/Eastern, Ben Hines wrote:
Use passive mode FTP transfers (to get through a firewall)? [Y/n]
'n' to this question and it fixed the problem. That seems
counter-intuitive.
curl -P is the command to disable PASV, and use PORT instead:
man curl "-P
I'm behind a NAT'ing non-proxying firewall and I could not update jikes
because the download failed. Specifically, this...
curl -f -L -s -S -O
ftp://www-126.ibm.com/pub/jikes/1.18/jikes-1.18.tar.bz2
... failed.
I modified NetAccess.pm such that the above became
curl -P - -f -L -s
On Sunday, Jan 12, 2003, at 11:42 US/Eastern, Jared wrote:
Frankly, because if you want a GUI solution, you should conform to the
way the GUI works. Creating something which is only "sorta like other
Mac apps" isn't good enough. You shouldn't expect others to work
around your decisions. Your de
On Sunday, Jan 12, 2003, at 11:22 US/Eastern, Ben Hines wrote:
On Sunday, January 12, 2003, at 08:03 AM, Jared wrote:
Using aliases is not Mac-like. Giving users the illusion of moving
their apps isn't good enough. Besides, at this point, anyone using
Fink isn't going to be stopped by having t
On Sunday, Jan 12, 2003, at 10:38 US/Eastern, Ben Hines wrote:
*IF* the user desires the flexibility to move these apps around, they
can install them without fink.
Or they can create a copy in their account or /Applications and do what
they want. But it'll be a copy no longer under Fink's contr
On Sunday, Jan 12, 2003, at 10:12 US/Eastern, Benjamin Reed wrote:
On Sunday, January 12, 2003, at 09:53 AM, Bill Bumgarner wrote:
Read the post. I didn't say that Apps installed by Fink should be in
/Applications. They should be in /sw/Applications, to further
follow the Fink
guide
On Sunday, Jan 12, 2003, at 07:49 US/Eastern, Max Horn wrote:
At 20:14 Uhr -0500 11.01.2003, Bill Bumgarner wrote:
On Wednesday, Jan 8, 2003, at 14:33 US/Eastern,
[EMAIL PROTECTED] wrote:
We don't generally accept bundled .apps in fink, because they can be
moved by the user, and also to
On Sat, 11 Jan 2003, Benjamin Reed wrote:
> > 2) Apps can't be moved any more readily than, say, /sw/bin/python or
> > /sw/share/doc/apache. Same goes for frameworks.
>
> What do you mean?
>
> You can pick up any application in /Applications and stick it somewhere
> else and it works just as
On Wednesday, Jan 8, 2003, at 14:33 US/Eastern,
[EMAIL PROTECTED] wrote:
We don't generally accept bundled .apps in fink, because they can be
moved by the user, and also to keep fink focused.
I know this has gone around before... but, to flog a dead horse a bit
more because this has really star
On Saturday, Dec 28, 2002, at 12:52 US/Eastern, Martin Costabel wrote:
Unfortunately, cvs does not have a cleanup option, "cvs update
--delete" or something. For a clean solution, one would have to write
a script that removes those files that are marked "?" by cvs update,
something equivalent t
On Saturday, Dec 28, 2002, at 06:13 US/Eastern, Martin Costabel wrote:
Bill Bumgarner wrote:
[]
Yes-- but the default for Fink is not to do that and I would bet that
most of the users that use selfupdate-cvs have no CLUE about
~/.cvsrc. The default for Fink should be to spew only the
On Friday, Dec 27, 2002, at 01:47 US/Eastern, Martin Costabel wrote:
Bill Bumgarner wrote:
This patch causes Fink to only output the files that have been added
or updated since the last fink update; all of the lines beginning
with '? ' are filtered out and all of the various
On Thursday, Dec 26, 2002, at 21:59 US/Eastern, Justin Hallett wrote:
I think we could just process it with perl and output what we want, I
have
done so in my bin dist sricts that way we can add something for C lines
like please fix this file, or something.
That'll require more work in compariso
This patch causes Fink to only output the files that have been added or
updated since the last fink update; all of the lines beginning with '?
' are filtered out and all of the various "updating directory" messages
from cvs are suppressed.
Index: perlmod/Fink/SelfUpdate.pm
... and the content is completely irrelevant to the Fink project.
Except that I borrowed very heavily from the PHP used to build the Fink
web site.
First, I wanted to say THANK YOU for putting together the Fink site --
it was very easy code to work with and the end result is about 100x
better
The xemacs-21.5.9-1 package is still broken [been so for 6 weeks or so]
-- it doesn't compile. In and of itself, it seems to be a minor
annoyance -- the last version has been working dandy for me [and
others] for quite a while.
Can/Should the package be rolled back to the last working version
I put together a python-nox-ssl package that builds python with SSL
support in the socket module; very useful for some development work
I'm doing.
Now, I want to do a fink update-all, but Fink is asking:
sudo /sw/bin/fink update-all
Password:
Information about 1596 packages read in 5 seconds
21.5.4 works correctly.
On Saturday, September 28, 2002, at 12:16 PM, jeff whitaker wrote:
> Bill: Several people have reported this with 21.5.9. Could you try
> 21.5.4 and let me know if it still happens?
>
> -Jeff
---
This sf.net email i
This occurred on 10.2 against the 10.2 version of Fink (I followed the
instructions to the letter and my installation is completely "normal").
I am only posting here because xemacs built fine up until now-- I'm
curious if something else may have broken or if this is a problem with
the new v
Whoah! That sounds more like patch is being called w/way too much
tolerance on the fuzz factor!
Usage: patch [OPTION]... [ORIGFILE [PATCHFILE]]
Input options:
...
-F LINES --fuzz LINES Set the fuzz factor to LINES for inexact
matching.
...
Fink should probably limit the fuzz to somethin
I have noticed on a number of packages -- including ones that I have
submitted -- that the fuzz factor on the patches has been drifting over
time. That is, as a version of a piece of software changes, the same
patch still works, but the exact location of the patch shifts slightly
and patch (t
autoconf is failing to build because it needs texi2html, but does not
depend or builddepend on it.
___
Sponsored by:
ThinkGeek at http://www.ThinkGeek.com/
___
Fink-devel mailing list
[EMAIL
when the user doesn't have the dev tools installed?
I just tried to do a 'fink selfupdate' on a system that only has the user
stuff installed (silly me-- i rebuilt the system at a whim and forgot to
install the dev tools as it is primarily a server).
The error message returned was not at
That would be the case. And brings up another issue. Should openldap-ssl
conflict/replace openldap?
On Saturday, May 25, 2002, at 10:07 PM, Benjamin Reed wrote:
> Bill Bumgarner [[EMAIL PROTECTED]] wrote:
>> /usr/bin/ld: Undefined symbols:
>> _ldap_p
cc -I/sw/include/db3 -fno-common -I../../include -I../../include
-no-cpp-precomp -I/sw/include -L/sw/src/openldap-ssl-2.0.23-3/openldap-2.0.
23/libraries -L/sw/lib -o slapd main.o daemon.o connection.o search.o
filter.o add.o charray.o attr.o entry.o config.o backend.o result.o
operation.o dn.o
n Sunday, April 28, 2002, at 12:01 PM, Max Horn wrote:
> At 11:44 Uhr -0400 28.04.2002, Bill Bumgarner wrote:
>>
>> - skip things that have already been downloaded
>
> fink fetch-missing does that.
Doh! Missed that-- and I even read the docs and everything. Cool.
&g
I'll give it a try, but it looks like I'll have to patch the whole
downloading thing fairly heavily if I am to make it such that it runs
without human interaction. I figure I'll need to:
- skip things that have already been downloaded
- only download those packages that change
...be to rip out all of the Fink code related to downloading source
tarballs and make it portable to multiple platforms?
I ask because I have a Linux box on the net that is doing very little
[lite mail services, that's it] and I would like to be able to point about
a dozen isolated Fink instal
If anyone from the Fink development community will be attending WWDC, I
would be very happy to buy you a beer or two -- if that's your cup of tea
-- as a way of saying THANK YOU for producing the incredible tool that is
Fink. The quality is incredible, the documentation easily outclasses
mos
I'm in the process of packaging Jython-- a pure java implementation of the
python language that we are considering for unit testing WebObjects
applications-- and have run across two issues that I thought I would run
by the community (to see if I can still come up with a compliant package):
-
If you want to see something truly frightening do a 'strings' on
/System/Library/PrivateFrameworks/Admin.framework/Resources/writeconfig.
Whoah! Therein lies dragons.
In any case, replacing the system provided apachectl with the one from
Fink causes the start/stop button in sharing to s
packages are ported across that use Apache
as a sort of desktop server-- this problem will become more and more
evident.
b.bum
On Wednesday, March 27, 2002, at 05:29 PM, Finlay Dobbie wrote:
>
> On Wednesday, March 27, 2002, at 09:15 PM, Bill Bumgarner wrote:
>
>> The fi
A conundrum...
How do you define the finkinfo for an apache module?
In particular, it wants to invoke apxs to (a) do the installation and
(b) adjust the configuration files as appropriate.
However, it assumes everything is already installed in the appropriate
location
There are two pieces; the start/stop functionality triggered by the
button in the Sharing prefs pane and the actual configuration.
The button works as you describe, I believe-- just a wrapper around
apachectl.
The configuration piece is more complex, but not hugely so. Look at the
configura
The fink installed apache doesn't support OS X's out of the box 'Web
Sharing' feature. This is problematic.
It means that if one wants to use Fink to install various apache related
features, one loses a significant chunk of OS X related functionality.
b.bum
Given that apache supports an include directive in the configuration
maybe a simple 'include /sw/etc/httpd/httpd.conf' would work for modifying
the system installed apache?
An advantage of possibly going this route is that it would allow Fink to
ADD modules to OS X Server's apache installa
Is it possible / how hard / why isn't...
... why isn't there a system-apache?It would seem to be possible save
for the missing header files [that can be obtained from the darwin source]
?
I ask specifically in the context of WebObjects -- Apple supplies a
compiled apache module webobjects
How about adding a feature to CustomMirror that indicates that the mirrors
should *always* be tried in the order they are mentioned in the .info file?
This would trivially fix the rsync related issue.
In the same vein, an option to cause Fink to automatically try the next
mirror until all m
Anyone?
___
Fink-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fink-devel
OK-- that almost works. Two problems (maybe):
(1) Fink seems to select between the mirrors at random [makes sense], but
I really want it to go in order so that it will generally succeed except
in those [hopefully rare] situations where the fink package description
has fallen behind the rsync
Rsync's version has been bumped frequently over the last few months; from
2.5.0 to 2.5.4 happened in just a few+ months w/a revision released every
few weeks.
While the current release is available via:
http://samba.anu.edu.au/ftp/rsync/
While old versions are always moved into:
Oh-- OK-- I see the problem. The patch file supplied with wxpython-wxgtk
makes the assumption that the glib include path is
@PREFIX@/lib/glib/include.
+includes = ['src',
+ '/usr/X11R6/include',
+ '@PREFIX@/lib/glib/include']
It looks like it could be patched t
[localhost:~] bbum% dpkg -S glib.h
glib: /sw/include/glib-1.2/glib.h
cc -DNDEBUG -g -O3 -Wall -Wstrict-prototypes -no-cpp-precomp -DSWIG_GLOBAL
-DHAVE_CONFIG_H -DWXP_USE_THREAD=1 -Isrc -I/usr/X11R6/include
-I/sw/lib/glib/include -I/sw/include/python2.2 -c src/helpers.cpp -o build/
temp.darwin-5
No prior WX packages installed-- appears that wxgtk installed fine.
copying wxPython/lib/PyCrust/version.py ->
build/lib.darwin-5.3-PowerMacintosh-2.2/wxPython/lib/PyCrust
running build_ext
building 'wxc' extension
creating build/temp.darwin-5.3-PowerMacintosh-2.2
creating build/temp.darwin-5.3
Any effort is better than none... if a quick fix works, go for it.
I wish I had more time to focus on this (and had the experience with PHP
to do so effectively)
On Friday, March 8, 2002, at 04:17 PM, David R. Morrison wrote:
>>> I'll take a look at updating it, if someone is willing to test w
I looked at it very briefly and it appears that the package is old enough
that it may be more productive to derive a new package based on a
combination of the old and on Scott Anguish's PHP articles at stepwise.com.
b.bum
On Friday, March 8, 2002, at 04:09 PM, David R. Morrison wrote:
> PHP w
is behind?
b.bum
On Friday, March 8, 2002, at 03:10 PM, Finlay Dobbie wrote:
>
> On Friday, March 8, 2002, at 07:13 PM, Bill Bumgarner wrote:
>
>> OpenOSX is shipping PHP 4.1.2, which fixes a number of bugs over 4.0.6--
>> the current version in Fink.
>
> Your point? Oh,
OpenOSX is shipping PHP 4.1.2, which fixes a number of bugs over 4.0.6--
the current version in Fink.
___
Fink-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fink-devel
The postfix package is broken; seems the url to the snapshot that was
being used is broken.
b.bum
___
Fink-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fink-devel
http://aspn.activestate.com/ASPN/Mail/Message/pythonmac-sig/1042874
It seems Fink is being used by more and more folks that are not
necessarily familiar with Darwin or Unix. Providing them with better
feedback during the installation process would likely make these folks'
lives easier.
b.bum
To check your inodes (assumes fileutils installed-- if not, drop the 'h'
flag). This is a default install a 20GB TiBook 500 w/OSX and OS9 in a
single partition.
[localhost:~] bbum% df -ih
FilesystemInodes IUsed IFree IUse% Mounted on
/dev/disk0s94.7M3.0M1.6M
Did you build the client or the server?
The client is trivial; pretty much builds out of the box.
The server is more problematic in that it requires the mod_dav.h header,
db4, and apr-- apache portable runtime.
mod_dav.h is interesting in that it can only be had by installing mod_dav.
Ap
I'm in the midst of porting subversion to OS X, creating a Fink package in
the process.
Subversion's source tree is a rather complex beast that includes a number
of other packages directly within and, hence, has a number of different
beasts that use various forms of autoconf.
It appears that
How about if FetchAltDir simply took an URL, as well as interpreting a
non-prefix'd path as a simple reference to a location in the filesystem?
b.bum
___
Fink-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fink-deve
It isn't necessarily local to my site; I might set up such a repository
of sourceballs on a machine @codefab or on my machine at home (cable modem)
and then point a small community of folks to it
On Wednesday, January 30, 2002, at 03:38 PM, Max Horn wrote:
> Well, if it is local to your
I probably have close to a dozen machines relatively 'local' to me that
all have different incarnations of Fink installed. Quite a number of them
are maintained be separate folk and are configured differently.
How about a 'source override' feature that allowed on to point Fink to a
central re
curl -L -s -S -O ftp://ftp.us.xemacs.org/pub/xemacs/xemacs-21.5/xemacs-21.
5.4.tar.gz
curl: (10) the username and/or the password are incorrect
### curl failed, exit code 10
Downloading the file "xemacs-21.5.4.tar.gz" failed.
(1) Give up
(2) Retry the same mirror
(3) Retry another mirror from
... consistently causing problems during the Fink update. Every time it
is updated, I have to hit at least 3 or 4 mirrors before one either
recognizes the package, allows the download, or actually starts the
download.
Automating the 'retry from another mirror until successful or all mirrors
Like I said, I have no opinion on the matter... merely throwing out a bit
of information from which others can derive the decision.
Personally, I don't care what packaging format is used, as long as it
works.
b.bum
___
Fink-devel mailing list
[EMAI
The linux kernel archives are offline due to a hardware failure...
attempting to build the man package is failing.
Yes, it'll come back someday.
But this raises the question:
Do we want to look into building a network of mirrors for the various
source bits in Fink?
I could do
I saw a conversation go by that mentioned the possibility of using an
alternative to dpkg for package management...
I have no particular preference but thought I would toss out that Apple is
moving more and more to using RPM internally and that the OpenPKG
project-- a sort of cross platform F
If you check out the source to SSHPassKey and the Tools/ directory from
mosxland.sourceforge.net, the build script can be used to automatically
package PB projects.
[~/bbum-developer/WOUnitTest] bbum% cd ../sourceforge/mosxland/
[~/bbum-developer/sourceforge/mosxland] bbum% Tools/build -name
'
68 matches
Mail list logo
