command line package manager for Osgeo4W
Home Page: http://trac.osgeo.org/osgeo4w/wiki/pkg-apt
License: GNU General Public License v2.0
I believe this will enable:
I released a new apt.exe testing package today. When I started setting up an environment to test what it will do in an upgrade situation I realised I need to put Osgeo4W into an older state, which means installing no-longer-current packages first. This isn't so easy.
I think the easiest approach is likely to be to adapting osgeo4w-regen.sh and genini to generate a setup.ini based on a date stamp. Then issuing an instruction like regen.sh 2014-06-01 would crawl the release package tree, ignore any files newer than said date, and generate setup_2014-06-01.ini.
Actually this is something useful for setup.exe too. I'm going to create a ticket for it upstream.
B:\code\apt>apt upgrade all
warning: missing packages: qgis tcltk
to install: qgis tcltk qgis-common
qgis-grass-plugin = d:/temp/o4w-cache/setup/http%3a%2f%2fdownload.osgeo.org%2
gin/qgis-grass-plugin-2.6.0-1.tar.bz2
remote: 22a54f29e679bb8fe2b837276152ddb7 qgis-grass-plugin-2.6.0-1.tar.bz2
local: 22a54f29e679bb8fe2b837276152ddb7 qgis-grass-plugin-2.6.0-1.tar.bz2
qgis-grass-plugin = d:/temp/o4w-cache/setup/http%3a%2f%2fdownload.osgeo.org%2
gin/qgis-grass-plugin-2.6.0-1.tar.bz2
remote: 22a54f29e679bb8fe2b837276152ddb7 qgis-grass-plugin-2.6.0-1.tar.bz2
local: 22a54f29e679bb8fe2b837276152ddb7 qgis-grass-plugin-2.6.0-1.tar.bz2
qgis-grass-plugin = d:/temp/o4w-cache/setup/http%3a%2f%2fdownload.osgeo.org%2
gin/qgis-grass-plugin-2.6.0-1.tar.bz2
remote: 22a54f29e679bb8fe2b837276152ddb7 qgis-grass-plugin-2.6.0-1.tar.bz2
local: 22a54f29e679bb8fe2b837276152ddb7 qgis-grass-plugin-2.6.0-1.tar.bz2
installing qgis 2.6.0-1
Traceback (most recent call last):
File "apt.py", line 1384, in <module>
__main__.__dict__[command] (packages)
File "apt.py", line 522, in upgrade
install(packages)
File "apt.py", line 229, in install
install_next(missing.keys(), set([]), set([]))
File "apt.py", line 258, in install_next
do_install(packagename)
File "apt.py", line 627, in do_install
pipe = tarfile.open(ball,'r')
File "B:\o4w\apps\Python27\lib\tarfile.py", line 1660, in open
return func(name, "r", fileobj, **kwargs)
File "B:\o4w\apps\Python27\lib\tarfile.py", line 1722, in gzopen
fileobj = bltn_open(name, mode + "b")
IOError: [Errno 2] No such file or directory: 'd:/temp/o4w-cache/setup/http%3
release/qgis/qgis/qgis-2.6.0-1.tar.bz2'
lately i came back to an old behavior of apt, that getopt (sometimes?) would not accept long name options like --ini and so on.
It's just a missing comma in the long name list, after 'root='
(options, params) = getopt.getopt (sys.argv[1:],
'cdhi:m:r:t:s:xv',
('cache=', 'download', 'help', 'mirror=', 'root='
'ini=', 't=', 'start-menu=', 'no-deps', 'debug', 'verbose'))May be you want to change this ...
B:\code\apt>apt --ini b:\o4w-packages\setup_2014-12-01.ini available
GetoptError: option --ini not recognized
B:\code\apt>apt -i b:\o4w-packages\setup_2014-12-01.ini available
IOError: [Errno url error] unknown url type: 'b'
B:\code\apt>apt -i file://b:\o4w-packages\setup_2014-12-01.ini available
IOError: [Errno socket error] [Errno 11004] getaddrinfo failed
From fix f3540a1
Today I noticed a knock on effect: even though the readonly file is removed from the file system and the installed-files-list successfully, that removal is not reported to console:
B:\code\apt>attrib b:\o4w\bin\iconv.dll
A B:\o4w\bin\iconv.dll
B:\code\apt>attrib +R b:\o4w\bin\iconv.dll
B:\code\apt>attrib b:\o4w\bin\iconv.dll
A R B:\o4w\bin\iconv.dll
B:\code\apt>apt remove iconv
iconv
removing iconv 1.9.1-1
removed: B:\o4w\include\iconv.h
removed: B:\o4w\lib\iconv.lib
B:\code\apt>attrib b:\o4w\bin\iconv.dll
File not found - B:\o4w\bin\iconv.dll
Would be nice if apt --root={...} --bits=64 setup saved the OSGEO4W_ROOT var in the shell so that we can just carry on with the next steps in the same session. Otherwise we have to keep re-using --root in every invocation which is a pain, leading to using set osgeo... at the beginning of a session, which leads to the --root parameter being not used, which is another way of saying "essentially useless". ;-)
B:\code\apt>apt update
Fetching http://download.osgeo.org/osgeo4w//setup.ini.bz2
...100%
Traceback (most recent call last):
File "apt.py", line 1341, in <module>
update ()
File "apt.py", line 479, in update
raise IOError('\n*** Error decompressing: %s' % archive)
IOError:
*** Error decompressing: d:/temp/o4w-cache/setup/http%3a%2f%2fdownload.osgeo.org%2fosgeo4w%2fsetup.ini.bz2
is not considered. Could be done in check_env(), would be a good thing if doing a apt setup and no environment variable has been set yet. Anyway, it exists, so it should do something.
X:> apt list
"list" not understood, please run "apt help"
Something broke recently. All other commands are ok.
Never mind. Forgot there is no list command, though now I can't remember the reason. Because of collision with python list() ? Might be sufficient to add an alias, similar to uninstall = remove.
With 32fbd7b we have a working info command, and a get_info() function which builds a dictionary of useful attributes. Now we have the necessary plumbing to rework all the functions which go out one by one to fetch and build the attributes one at a time into simple lookups from the dict.
So instead of calling on 3 chained functions
def ball(packages):
'''Print full local path name of package archive '''
print "\n%s = %s" % (p, get_ball(p))
def get_ball (packagename):
url, md5 = get_url (packagename)
return '%s/%s' % (downloads, url)
def get_url (packagename):
if not dists[distname].has_key (packagename) \
...~20 lines snipped...
return filename, md5
we can just do
pkg_info = get_info(pkg)
print pkg_info['ball']
A delightful collapse in complexity which should make understanding and working on this program a whole lot easier and more enjoyable!
Before going to deep into touching all the info-like functions we should spend some time thinking about building a nested list/dictionary one level higher. Get_info() gives us a dict for each package, one at a time. Should we also build a list-of-pkg-dicts for everything all in one go?
B:\code\apt>apt remove iconv
iconv
warning: iconv not installed
B:\code\apt>apt install iconv
Already installed!
KeyError: 'install'
B:\code\apt>apt --debug install iconv
### DEBUG: parse_setuprc ###
last-cache: B:/o4w//var/cache/setup
last-mirror: http://download.osgeo.org/osgeo4w/
### DEBUG: <module> ###
last-mirror: http://download.osgeo.org/osgeo4w/
last-cache: B:/o4w//var/cache/setup
Already installed!
iconv 1.9.1-1
### DEBUG: write_setuprc ###
Wrote B:/o4w//etc/setup/setup.rc
3 bugs here:
--debug changes the behaviour, 2nd bug doesn't manifest. Using debug should not change behaviour.In b3de40f (and everywhere?) stdout isn't being captured.
Running test-architecture.bat:
Couldn't open C:/Users/Matt/AppData/Local/Temp/apt-9460/etc/setup//setup.rc, setting empty
Fetching http://download.osgeo.org/osgeo4w//x86_64/setup.ini
Skipping download - url modified time isn't newer than local file
D:\Public\Downloads/http%3a%2f%2fdownload.osgeo.org%2fosgeo4w%2f\x86_64/setup.ini
Osgeo4w folders and setup config exist; skeleton environment is complete.
You might try `apt available` and `apt install` next.
apt---bits=64 setup.stdout - nothing, doesn't exist!
apt---bits=64 setup.stderr
Root dir not found, creating C:/Users/Matt/AppData/Local/Temp/apt-17074
creating C:/Users/Matt/AppData/Local/Temp/apt-17074/etc/setup/
creating C:/Users/Matt/AppData/Local/Temp/apt-17074/etc/setup//installed.db
getting C:/Users/Matt/AppData/Local/Temp/apt-17074/etc/setup//setup.ini
B:\code\apt\tests>apt --bits=64 setup
Couldn't open d:/temp/apt-mar31xx/etc/setup//setup.rc, setting empty
Root dir not found, creating d:/temp/apt-mar31xx
creating d:/temp/apt-mar31xx/etc/setup/
creating d:/temp/apt-mar31xx/etc/setup//installed.db
getting d:/temp/apt-mar31xx/etc/setup//setup.ini
Fetching http://download.osgeo.org/osgeo4w//x86_64/setup.ini
Skipping download - url modified time isn't newer than local file
D:\Public\Downloads/http%3a%2f%2fdownload.osgeo.org%2fosgeo4w%2f\x86_64/setup.ini
Osgeo4w folders and setup config exist; skeleton environment is complete.
You might try `apt available` and `apt install` next.
Occasionally a file in the downloads cache is corrupted or doesn't match it's mirror equivalent. Currently one has to remove that local file manually, and locating it in order to do so is a pain. We should either automatically overwrite the local file when downloading, or provide a means to clean the cache of that and/or all files.
...might be able to join or match this with removing versions older than X.
D:\>pip install --editable b:\code\apt
Obtaining file:///b:/code/apt
Running setup.py (path:b:/code/apt\setup.py) egg_info for package from file://
/b:/code/apt
Installing collected packages: apt
Running setup.py develop for apt
B:\o4w\apps\Python27\lib\site-packages\setuptools-1.3.2-py2.7.egg\setuptools
\command\easy_install.py:320: UserWarning: Unbuilt egg for apt-Osgeo4w-package-m
anager [unknown version] (b:\code\apt)
Creating b:\o4w\apps\python27\lib\site-packages\apt.egg-link (link to .)
Adding apt 0.3 to easy-install.pth file
Installing apt-script.py script to B:\o4w\apps\Python27\Scripts
Installing apt.exe script to B:\o4w\apps\Python27\Scripts
Installing apt.exe.manifest script to B:\o4w\apps\Python27\Scripts
Installed b:\code\apt
Successfully installed apt
Cleaning up...
D:\>apt
Traceback (most recent call last):
File "B:\o4w\apps\Python27\Scripts\apt-script.py", line 9, in <module>
load_entry_point('apt==0.3', 'console_scripts', 'apt')()
File "b:\code\apt\apt.py", line 1271, in main
last_mirror = get_config('last-mirror')
File "b:\code\apt\apt.py", line 727, in get_config
f = os.path.join(config, fname)
NameError: global name 'config' is not defined
My last commit on amr66/apt now allows to download tinyows and dependencies, but it fails with apache. I got a message from shlex called by get_menu_links(). I didn't try much further, but it seems to be a link in apache.bat.done that causes an exception on link = shlex.split(line)[1]. Error is "Value Error: No closing quatation". Line from bat is:
xxmklink "%OSGEO4W_STARTMENU%\Apache\OSGEO4W-Apache-Monitor.lnk" %OSGEO4W_ROOT%\Apache\bin\ApacheMonitor.exe"
Is this a " which is missing in front of %OSGEO4W_ROOT%?
PS:
I just tried to install with setup exe: It works, even the start menu entries are there,
so by the way: why do we do this: parsing a batch file that has already finished it's work? What intention is behind get_menu_entries?
It's been a very long time since I've let anyone know about what's going in apt world, and all of the main commands now do useful work. It's time to update the main Trac instance, and put in a pointer in SVN that points here.
This is worth doing, even if there isn't anything truly new in what apt does (no new commands or features), mostly better screen reporting, but there are some worthwhile bug fixes too (IMO).
Recap from 881fe03:
--arch= [x86 | x86_64] 8c36813) (or --bits= [32 | 64] 881fe03)apt, apt-x64 (though it would be identical script inside)2 problems here:
B:\code\apt>apt download gdal-filegdb
Fetching http://download.osgeo.org/osgeo4w/x86/x86/release/gdal/gdal-filegdb/gdal-filegdb-1.11.1-4.tar.bz2
...100%
gdal-filegdb = d:/temp/o4w-cache/setup/http%3a%2f%2fdownload.osgeo.org%2fosgeo4w%2fx86/x86/release/gdal/gdal-filegdb/gda
l-filegdb-1.11.1-4.tar.bz2
remote: b434e716cc2ff6a0eb252fd725219d20 gdal-filegdb-1.11.1-4.tar.bz2
local: 81802d554670b1cb3d40c770ecdcdff5 gdal-filegdb-1.11.1-4.tar.bz2
...snip...
contents of the local file:
<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
<html><head>
<title>404 Not Found</title>
</head><body>
<h1>Not Found</h1>
<p>The requested URL /osgeo4w/x86/x86/release/gdal/gdal-filegdb/gdal-filegdb-1.11.1-4.tar.bz2 was not found on this server.</p>
<hr>
<address>Apache/2.2.16 (Debian) Server at download.osgeo.org Port 80</address>
</body></html>
As @amr66 noted: Comparing version strings fails by simple comparison with logical less than (<), 1.11.1 < 1.9.1 gives true
Resources
Package Installed Available
----------------- --------- ----------
TypeError: no_package() takes at least 2 arguments (0 given)
Hi Matt,
i have to correct string mirror in apt.py to successfully run apt setup/update. I added arch part /x86/ (in line 530, get_mirror () ). Also there is no 'http://download.osgeo.org/osgeo4w/setup.ini.bz2'.
What am i missing?
apt setup -> craches without a correct url
But adding x86 to mirror string sure has side effects?
Regards,
Andreas
apt info [package] should report on whether the package is installed or not.
I Tried running python program to controlling thorlabs motor controller. I found error codes such below:
Traceback (most recent call last):
File "C:\Users\N 417 TA\Desktop\meanshift.py", line 6, in
import thorlabs_apt as apt
File "C:\ProgramData\Anaconda3\lib\site-packages\thorlabs_apt-0.1-py3.6.egg\thorlabs_apt_init_.py", line 1, in
from .core import *
File "C:\ProgramData\Anaconda3\lib\site-packages\thorlabs_apt-0.1-py3.6.egg\thorlabs_apt\core.py", line 1373, in
_lib = _load_library()
File "C:\ProgramData\Anaconda3\lib\site-packages\thorlabs_apt-0.1-py3.6.egg\thorlabs_apt\core.py", line 1354, in _load_library
filename = ctypes.util.find_library("APT")
AttributeError: module 'ctypes' has no attribute 'util'
As information, I using python 3.6.0 64 bit.
yukyukyukyuk. No wonder my attempts to create a unique list with
x = list(setup(some_list))
was returning NONE all the time!!
Using pyinstaller with out of the box settings works, but the resulting file is really big: 9mb when bundled as onefile, and 20mb when as onedir (!)
Inspecting the expanded folder I see Tk and Tcl are included. We certainly don't need those modules, and I suspect there are probably others that we can exclude too.
B:\code\apt>apt -i setup_test.ini install qgis
PKGS: Checking install status: qgis
Requirement Installed (Available)
-------------------------------------------
msvcrt 1.0.1-12 (1.0.1-12)
setup 1.0.7-10 (1.0.7-10)
...
python-dateutil 2.2-1 (2.2-1)
pytz 2014.2-1 (2014.2-1)
qgis
REQS: --- To install: qgis-common qgis
Preparing to download: qgis-common
WindowsError: [Error 5] Access is denied: 'C:\\Users\\Public'
B:\code\apt>apt upgrade gdal-filegdb
to install: gdal-filegdb
Fetching http://download.osgeo.org/osgeo4w/x86/x86/release/gdal/gdal-filegdb/gdal-filegdb-1.11.1-4.tar.bz2
...100%
gdal-filegdb = d:/temp/o4w-cache/setup/http%3a%2f%2fdownload.osgeo.org%2fosgeo4w%2fx86/x86/release/gdal/gdal-filegdb/gda
l-filegdb-1.11.1-4.tar.bz2
remote: b434e716cc2ff6a0eb252fd725219d20 gdal-filegdb-1.11.1-4.tar.bz2
local: 81802d554670b1cb3d40c770ecdcdff5 gdal-filegdb-1.11.1-4.tar.bz2
Traceback (most recent call last):
File "apt.py", line 1385, in <module>
__main__.__dict__[command] (packages)
File "apt.py", line 522, in upgrade
install(packages)
File "apt.py", line 224, in install
download(p)
File "apt.py", line 169, in download
md5(p)
File "apt.py", line 318, in md5
raise TypeError('file md5 does not match for ' + ball)
TypeError: file md5 does not match for gdal-filegdb-1.11.1-4.tar.bz2
Apt doesn't know about the %OSGEO4W_MENU_LINKS% and %OSGEO4W_DESKTOP_LINKS% variables introduced in setup.exe v1.0.6-5.
With the qgis-dev package this manifests with a mkdir was unexpected at this time. error message at the console during postinstall phase, consequently the Start Menu and Desktop shortcuts are not created, nor are the relevant registry entries created/updated.
B:\code\apt>apt -i setup_test.ini install apt-exe
PKGS: Checking install status: apt-exe
Requirement Installed (Available)
-------------------------------------------
apt-lib 0.3-3 (0.3-3)
apt-exe - (0.3-3)
REQS: --- To install: apt-exe
Preparing to download: apt-exe
--- Verifying local file's md5 hash matches mirror
True
remote: 334808074060560b8e8d78a86a4fa1e1
local: 334808074060560b8e8d78a86a4fa1e1
Skipping download of apt-exe-0.3-3.tar.bz2, exists in cache
apt-exe = C:\Users\Public\Downloads\OSGeo4W-setup-cache\http
-exe\apt-exe-0.3-3.tar.bz2
--- Verifying local file's md5 hash matches mirror
True
remote: 334808074060560b8e8d78a86a4fa1e1
local: 334808074060560b8e8d78a86a4fa1e1
Reports md5 in package archive (334808074060560b8e8d78a86a4fa1e1) matches md5 from setup.ini , but that's not the case (a662af2ce6b94ffa011a5252c3a3cd0c):
@ apt-exe
sdesc: "Command line OSGeo4W Installer/Updater/Remover"
ldesc: "Command line OSGeo4W Installer/Updater/Remover - exe version (doesn't require python)"
category: Commandline_Utilities
requires: apt-lib
version: 0.3-3
install: x86/release/apt/apt-exe/apt-exe-0.3-3.tar.bz2 1963823 a662af2ce6b94ffa011a5252c3a3cd0c
[prev]
version: 0.3-2
install: x86/release/apt/apt-exe/apt-exe-0.3-2.tar.bz2 1963590 734a96cd949b1ecf661047c9ec6adc0c
It appears apt is ignoring the custom ini and reverting to the default for md5 check. Apt does however use the custom ini when installing (apt -i setup_test.ini install foobaz).
:-/
It would be much more useful if apt requires looked in both directions and reported what package X requires AND what requires package X.
Curious, apt upgrade pyqt4 failed, in the sense that it decided there was nothing to do, while apt upgrade gpsbabel and another package which I don't remember, worked just fine. Workaround to just remove the pyqt4 package and then install/upgrade it again was quick and easy, but still should have worked in the first place.
B:\code\apt>apt new
*** Warning: 'grass71-dev' is missing 'install' entry in setup.ini. This might cause problems.
Package Installed Available
----------------- --------- ----------
gdal 1.11.1-5 (1.11.2-1)
gdal-filegdb 1.11.1-5 (1.11.2-1)
gdal-python 1.11.1-5 (1.11.2-1)
gpsbabel 1.4.2-1 (1.4.4-1)
grass 6.4.4-1 (7.0.0-1)
osg-libs 3.0.0-1 (3.1.7-1)
osgearth-libs 2.0.0-1 (2.4-1)
pyqt4 4.8.3-2 (4.11.3-1)
python-numpy 2.7.1.5.1-1 (2.7.1.7.0-1)
python-qscintilla 2.6.2-2 (2.8.4-1)
qgis 2.6.0-1 (2.8.1-1)
qgis-common 2.4.0-1 (2.8.1-1)
qgis-dev 2.7.0-21 (2.9.0-34)
qgis-grass-plugin 2.4.0-1 (2.8.1-1)
qscintilla 2.6.2-2 (2.8.4-1)
qt4-libs 4.7.1-6 (4.8.6-1)
B:\code\apt>apt upgrade pyqt4
*** Warning: 'grass71-dev' is missing 'install' entry in setup.ini. This might cause problems.
+++ get_requires() is depcrecated. Please use get_all_dependencies().
PKGS: Checking install status: pyqt4
pyqt4 - True
REQS: Checking dependencies installed: python-tools qt4-libs shell python-testsuite python setup zlib openssl python-cor
e python-help python-tcltk sqlite3 msvcrt
python-tools - True
qt4-libs - True
shell - True
python-testsuite - True
python - True
setup - True
zlib - True
openssl - True
python-core - True
python-help - True
python-tcltk - True
sqlite3 - True
msvcrt - True
Packages and required dependencies are installed.
pyqt4 4.11.3-1
python-tools 2.7.4-1
qt4-libs 4.8.6-1
shell 1.0.0-13
python-testsuite 2.7.4-1
python 2.7.4-2
setup 1.0.7-10
zlib 1.2.3-1
openssl 1.0.1g-1
python-core 2.7.4-3
python-help 2.7.4-1
python-tcltk 2.7.4-1
sqlite3 3.7.17-1
msvcrt 1.0.1-12
Took me some nerves, because i couldn't build a performing exe. Always got an error, that a sub-package of requests was missing. The same is described here: https://github.com/kennethreitz/requests/issues/2465
So i went back to request 2.5.1, using pip install requests==2.5.1
At least no bug - but we should consider and look out for a fix.
There is a function md5 and a global variable md5. One of them has to go.
Currently all my testing is done by running in a command shell and then comparing against my memory of what to expect. This is inadequate.
I've avoided unittest to this point because in my (limited) experience with it I spend more time debugging the tests themselves that the functions they're supposed to be verifying. ...
Still, something needs to be done. :)
See 0f2a644 for an example
Setup.exe creates this, we don't need it so I don't see much value in maintaining feature parity in this particular instance. Presumably it is to avoid downloading a new package list on every invocation, only when older than X. Apt doesn't need this since it has it's own apt update command that's fired at user discretion.
It could be helpful though, no point in having Setup download a new package ini if apt did it just a few minutes ago, but I'm not going to implement it myself while there's bigger things to tackle. Pull requests welcome.
apt search only looks in package names. It would be much more useful it looked in description fields as well (or perhaps any part of the .ini record).
Spent some time researching how to search nested dictionaries, but didn't come up with anything that I both (a) understood, and (b) didn't involve a lot of extra work.
As explored in #40, the intermingling of Apt progress reporting and the messages coming from batch files extracted from packages can make it difficult to determine where a problem is coming from. It would be good to portray these messages and errors in way that makes it easier to differentiate.
One idea is to redirect batch stdout stderr to another console or log file(s). The more I think about this though the less attracted I am. Who wants to wade through possibly dozens or hundreds of logfiles in the event of an issue? Maybe could just draw them in different colours?
Anyway, no solution or plan for one at this point, just a ticket to attach one to later.
B:\code\apt>apt --root=%temp%\%random% -b=64 setup
Couldn't open C:\Users\Matt\AppData\Local\Temp\5024/etc/setup//setup.rc, setting empty
error: `--bits` is required parameter for Setup
*** CPU Architecture not defined. Please use `--bits [32 | 64]`
When I build the sphinx docs locally with make html there is a module page populated with content, but at http://apt.readthedocs.org/en/latest/apt.html the page is empty.
Under Builds, https://readthedocs.org/builds/apt/2900858/, there is an error with ctypes.
/var/build/user_builds/apt/checkouts/latest/docs/source/apt.rst:4: WARNING: autodoc: failed to import module u'apt'; the following exception was raised:
Traceback (most recent call last):
File "/home/docs/checkouts/readthedocs.org/user_builds/apt/envs/latest/local/lib/python2.7/site-packages/sphinx/ext/autodoc.py", line 385, in import_object
__import__(self.modname)
File "/var/build/user_builds/apt/checkouts/latest/apt.py", line 38, in <module>
import knownpaths # for fetching Windows special folders
File "/var/build/user_builds/apt/checkouts/latest/knownpaths.py", line 5, in <module>
from ctypes import windll, wintypes
File "/usr/lib/python2.7/ctypes/wintypes.py", line 23, in <module>
class VARIANT_BOOL(_SimpleCData):
ValueError: _type_ 'v' not supported
also: http://trac.osgeo.org/osgeo4w/ticket/283
D:\osgeo>apt install shell
to install: setup shell msvcrt
...downloads happen...
Traceback (most recent call last):
File "<string>", line 1524, in <module>
File "<string>", line 299, in install
File "<string>", line 316, in install_next
File "<string>", line 315, in install_next
Exception: Required package setup from shell is a circular reference with a prev
ious dependent
D:\osgeo>apt requires shell msvcrt setup
----- "shell" requires the following to work -----
msvcrt
setup
----- "msvcrt" requires the following to work -----
----- "setup" requires the following to work -----
msvcrt
shell
D:\osgeo>apt --no-deps install shell
to install: shell
...
installing shell 1.0.0-13
D:\osgeo\Osgeo4w-apt>mkdir "C:\ProgramData\Microsoft\Windows\Start Menu\Programs
\OSGeo4W"
Access is denied.
D:\osgeo\Osgeo4w-apt>xxmklink "C:\ProgramData\Microsoft\Windows\Start Menu\Progr
ams\OSGeo4W\OSGeo4W.lnk" "D:\osgeo\Osgeo4w-apt\OSGeo4W.bat" " " \ "OSGeo for Win
dows command shell" 1 "D:\osgeo\Osgeo4w-apt\OSGeo4W.ico"
'xxmklink' is not recognized as an internal or external command,
operable program or batch file.
...
Post_install complete, return code 1
Setup.exe has moved it's settings for last-mirror, last-cache, etc. into a single file: ./etc/setup/setup.rc. We need to do the same to maintain compatibility.
mirrors-lst
http://download.osgeo.org/osgeo4w/;OSGeo;USA;California
window-placement
44,0,0,0,0,0,0,0,1,0,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,151,1,0,0,185,0,0,0,185,3,0,0,72,2,0,0
last-mode
Advanced
last-mirror
http://download.osgeo.org/osgeo4w/
net-method
Direct
last-cache
C:\Users\Matt\Downloads
last-menu-name
OSGeo4W_default
urllib is ancient, urllib2 is newer, but requests is clearly the favourite. I think this would be a good move, even if it is not included in the standard library.
Related discussion #26
With the last update() of setup.ini i recognized that grass71-devel has no entry for install.
So in parse_setup_ini() this will cause an KeyError Exception:
d['zip_path'],d['zip_size'],d['md5'] = d['install'].split()
you could change it to
try:
d['zip_path'],d['zip_size'],d['md5'] = d['install'].split()
except KeyError:
print "warning: ", d['name'], "has no entry install"
d['zip_path'],d['zip_size'],d['md5'] = ('','','')
But this causes errors later, if one tries to access that information, e.g. on download/install
I've not been using Raise or Exceptions properly. Some of that is inheritance from the original cyg-apt script, which just used a bare raise 'string message' syntax, and some of it is me, in ignorance, using TypeError all over the place, even when the error is not a type error. I've since learned enough to recognize the issue, but don't think I know enough about exception handling to solve it properly. We'll see I guess ;-)
python-core = C:\Users\Public\Downloads\http%3a%2f%2fdownload.osgeo.org%2fosgeo4w%2f\x86_64\release\python\python\python-core\python-core-2.7.5-3.tar.bz2
--- Verifying local file's md5 hash matches mirror
True
remote: c38a03b6ffa6b43fcf514fe5e1b7614c
local: c38a03b6ffa6b43fcf514fe5e1b7614c
IOError: [Errno 13] Permission denied: '.\\etc\\ini\\python-core.bat'
Failed to execute script apt
The work-around:
attrib -r C:\OSGeo4W64\etc\ini\python-core.bat
Because I don't understand the code in get_requires() i wrote a function for my own, but failed with the order of packages. There are indeed packages that need other packages to be already installed! 8-o
e.g. qgis depends on qgis-common which tries to execute a command %OSGEO4W_ROOT%\apps\qgis\crssync, an exe to do something with proj's spatial references. Anyway, if so, during install there must be a certain order. So i came back to apt's get_requires() and yes, the order is different to mine.
In setup.ini all packages have there "requires" and those can have their own "requires" too. I tried to recursively collect the required packages but ended in an infinite loop :-(
The reason is that some elements are "mutual depending", eg. "shell" needs "setup" and vice versa.
I still don't know, why and how get_requires() work, but it definitely does not follow the recursive way, even not in its output.
For example "install python"
A strict order would be (reading setup.ini carefully): 'setup', 'msvcrt', 'shell', 'sqlite3', 'python-core', 'python-help', 'python-tcltk', 'python-testsuite', 'python-tools', 'python'
but get_requires() gives 'shell', 'python-testsuite', 'python', 'setup', 'sqlite3', 'python-core', 'python-help', 'python-tcltk', 'python-tools', 'msvcrt'
All this works well, but what if a package-to-be must execute an python script which then depends on msvcrt?
It ended all up with my own get_dependencies(), where i tried to recursively, not falling into the infinite-loop-trap, sort packages according to dependence:
def get_dependencies(packages, nestedl, parent=None):
"""a recursive lookup for required packages in order of dependence"""
if isinstance(packages, basestring): packages = [packages]
for p in packages:
mm = get_requires(p)
if parent:
inspos = nestedl.index(parent)
nestedl.insert(inspos, p)
else:
nestedl.append(p)
delete_in_existing(mm, nestedl)
if mm:
nestedl = get_dependencies(mm, nestedl,p)
return nestedlAnd here is the driver:
dlist = []
dlist = get_dependencies(packages, dlist)
print dlistMay be, this could be a replacement for get_requires, even there's actually no need for it...
Any comment is welcome!
with get_all_dependencies we get all packages to be installed in an list, odered by dependencies, starting with basic packages and ending with those dependending on the foregoing. So we need not to distinguish between packages[] and requiered[] anymore. Also we clean the list froms doubles, so we needn't unique() twice?
just
When working on better information messages from apt info command I realized I've made a muddle:
4b9190f The main data tables need clear separation, "package names and version available on server" and "what is actually locally installed". The old functions actually maintained this better before I started messing with them. (In my defense, they were horribly opaque tho'.)
For instance I grew the idea that
apt ballreconstructed the version name from the currently available archive filename on the server. Yes, get_ball() constructs from the filename, but not from the filename in setup.ini, it's actually from the name in installed.db -- a crucial difference!TODO: rescue good parts of old functions, clearly separate local/server info, dump overlapping dict records.
So something like
SETUP = parse_setup_ini()
INSTALLED = parse_installed_db()
I want to think carefully about the names though. We'll be using them frequently. They should make intuitive sense when read, while being short enough to cause no typing strain and keep our code phrases short and fast to read.
ALLCAPS is out; no shouting.setup_ini is being used (but could be moved, with some effort)setup_d, installed_d ?sini_d, inst_d ?ini_d, inst_d ?S, I ? (is this still shouting?)A couple of spelled out phrases:
p = packagename
server_ver = sini_d[p]['version']
local_ver = inst_d[p]['version']
server_ver = S[p]['version']
local_ver = I[p]['version']
if sini_d[p]['version'] > inst_d[p]['version']:
do_upgrade(p)
if S[p]['version'] > I[p]['version']:
do_upgrade(p)
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.