Ok, the clipartbrowser cvs module is now synched up again with the latest tarballs, and the cannonical download location for releases is now
http://www.python.org/pypi/clipartbrowser
(it used to be on Berlios). If you download a release (they work better than ever :), install using "python setup.py install" (assuming you have the dependencies met). The "easy_install" method mentioned earlier works too.
Greg
On 9/20/05, Greg Steffensen <greg.steffensen@...400...> wrote:
I didn't explain the technical situtation with the install method very well. Basically, if I write a file "setup.py" that defines certain information, users should be able to execute "python setup.py install" to install the program correctly, and I should be able to execute commands "python setup.py bdist_win" to create actual win32 EXE files containing pretty automated installer GUIs. In addition, users will be able to install the way I described in the release email as well. In general, setup.pyuses the "setuptools" which are the successor to the "distutils", which is the distribution/installation tools included with python. I'm still not certain whether I should use the setuptools or the older distutils, and am figuring stuff like that out, but they're both supposed to be the python-friendly replacement for Makefiles, and I'm trying to reorganize the code to do stuff the "correct" way.
Greg
On 9/20/05, Greg Steffensen <greg.steffensen@...400...> wrote:
Hey, yeah, this is kinda what I was referring when I said that the packaging isn't as polished as it needs to be. For what its worth, most of the changes (bug fixes, performance improvements) actually are committed to CVS; I just didn't commit commit the change in the install method simply because I was having trouble figuring out how to get files like COPYING, NEWS, etc. included using that method, and I didn't want to put a totally messed up version into CVS. Basically, I actually really released more on a whim, not because I'd worked everything out, but because I'd gone too long without releasing code, wanted people to know that work was ongoing. So, at a minimum, I should have indicated that this was definitely a development release.
So, solutions... in terms of where formal releases (formal tarballs) should go, I actually think the python package index is definitely the correct place... its now the "official" place where the python folk reccomend that python projects be listed, and it will offer a number of technical advantages, including publication of new releases, and installation automation. I'm very happy keeping CVS at Inkscape though (not that PyPI offers CVS anyway), and treating the code as a subproject of Inkscape. But as long as it offers the ability to be installed standalone as well, I'd like to make standalone tarballs available from PyPI. In general, I'm learning more about python, and am trying to make my practices for installation, etc. match accepted practice.
I'll try to get CVS synched up with the current state of the project very soon; I'm still having some trouble with the python installation tools, and have sent an email to their mailing list for advice, but haven't heard back yet. Once I know how to get the documentation files included, and where to put the main install file (setup.py), I'll get that committed and CVS will contain the official version of the code gain. Again, I would have waited until this was done to release, but just felt bad about having gone too long without releasing already, and was overeager to get something out.
Anyway, does this cover the stuff you were concerned about? If not, let me know. Also, you may recall that someone posted a link to a mockup for a very similar project that they'd found, and I replied that this was "very depressing" :). I used that mockup to develop version 0.4, and emailed the designer (as the community suggested); he's finally written me back, and said he's interested in contributing, which rocks. Later,
Greg
On 9/19/05, Jon Phillips <jon@...235...> wrote:
<offlist />
On Mon, 2005-09-19 at 05:20 -0400, Greg Steffensen wrote:
I'm releasing another version of the clip art browser. Lots of bug fixes, some performance improvements, code cleanups, and a new installation procedure (yes, yet another).
Installation is now 3 steps, and hopefully easier than ever:
Satisfy the dependencies of Python 2.4 and PyGTK 2.6.
Download and run the following script, which installs the python
setuptools:
http://peak.telecommunity.com/dist/ez_setup.py
- Run "easy_install clipartbrowser"
As you can see, I've switched the installation system from Makefiles
to the python setuptools. The setuptools are still under development (they're apparently going to be included in python 2.5), and I'm
still
very much learning how to use them, so the packaging isn't as
polished
as it needs to be, but this makes installation a breeze. Eventually, it should allow arbitrary python dependencies to be intelligently installed as well using the same process.
I'm still working on getting this integrated back into the Inkscape effects menu, that's next on the list. As always, feedback greatly appreciated.
Greg,
I thought we were going to do the next release from Inkscape CVS? What you have done is a non-standard release of the project now that we consolidated the code into Inkscape.
Also, you have now used a non-standard approach to packaging which then again breaks what users and developers are use to.
The major thing though is that we were going to do a solid release which means you need to involve the people who would contribute to this in the release to help smooth out the bugs.
Instead, now you have released from another location thus confusing users and developers more. The problem with this is that it is not pro-community.
Anyhow, I'm curious what you think and also want to get these things on track with the community.
It is so great that you are rocking these changes and I'm so proud of what you are doing so I don't want to dismay you, but really I think the next release has to be really on track with the community and we should coordinate it, push it out through the sourceforge system we have in place, and push the press release globally.
Without these procedures, honestly, it is not very likely that others will help on development nor use your work.
So, lets talk some more and sort these things out.
Jon
-- Jon Phillips
San Francisco, CA USA PH 510.499.0894 jon@...235... http://www.rejon.org
MSN, AIM, Yahoo Chat: kidproto Jabber Chat: rejon@...896... IRC: rejon@...897...
Inkscape (http://inkscape.org) Open Clip Art Library ( www.openclipart.orghttp://www.openclipart.org )
I have been looking at some large SVG symbol files from IAN and having trouble with ungrouping them. I can get down to individual symbols and then I get stuck.
A typical case is an SVG image of a fish. This opens in Inkscape 0.42 readily and when selected shows "Group of 1 object". When I try and ungroup it it goes to "Clone of: Group of 1 object" and I don't get any further. Obviously I want to ungroup it to manipulate the image. It has colours, gradients and maybe transparencies.
I can save it as a png and use it that way but I want to retain and use it in vector format. Any suggestions on getting it ungrouped?
Erik
On 9/21/05, Erik <kaver@...68...> wrote:
A typical case is an SVG image of a fish. This opens in Inkscape 0.42 readily and when selected shows "Group of 1 object". When I try and ungroup it it goes to "Clone of: Group of 1 object" and I don't get any further. Obviously I want to ungroup it to manipulate the image. It has colours, gradients and maybe transparencies.
First, please always attach or give a link to the file you refer to.
Second, judging by your description, that object is a clone. Press Shift+D to select to its original. If it says the original is in defs and cannot be selected, then you can just unlink the clone by Shift+Alt+D. After that it's a regular object (in this case a group).
-- bulia byak Inkscape. Draw Freely. http://www.inkscape.org
Hi Erik, I just registered with IAN and downloaded the package. Some very nice stuff in there. First thing i noticed was the size of the SVGs, they've been saved from illustrator with what amounts to an embeded ai file in them, stored in the pgf node at the bottom of each doc. It would seem this is part of the reason they seem to behave so odd in IS. As an experiment i deleted the i:pgf from one of them in a text editor, saved it and when i opened it in inkscape it now ungrouped a lot more happily. Still behaves oddly though as the file makes extensive use of svg:symbol, which it would appear we dont have much of a ui for dealing with, they also use switch which I dont know what we do with. If you go into the xml editor, and find the symbol you want, you can move the group defining the geom from inside the symbol to within the root doc. It seems to rotate them by 180 as you do, but you can easily fix that with the selector tool bar. Once its out you can copy and paste as normal. Gone midnight here so I'm not gonna dig too far but it does raise some interesting questions about how we handle symbols and switch statements at the minute.
hope thats some help
John
btw, for the rest of you, IAN can be found here: http://ian.umces.edu/index.html?http&&&ian.umces.edu/conceptuald...
--- Erik <kaver@...68...> wrote:
I have been looking at some large SVG symbol files from IAN and having trouble with ungrouping them. I can get down to individual symbols and then I get stuck.
A typical case is an SVG image of a fish. This opens in Inkscape 0.42 readily and when selected shows "Group of 1 object". When I try and ungroup it it goes to "Clone of: Group of 1 object" and I don't get any further. Obviously I want to ungroup it to manipulate the image. It has colours, gradients and maybe transparencies.
I can save it as a png and use it that way but I want to retain and use it in vector format. Any suggestions on getting it ungrouped?
Erik
__________________________________ Yahoo! Mail - PC Magazine Editors' Choice 2005 http://mail.yahoo.com
On 9/21/05, Erik <kaver@...68...> wrote:
A typical case is an SVG image of a fish. This opens in Inkscape 0.42 readily and when selected shows "Group of 1 object". When I try and
ungroup
it it goes to "Clone of: Group of 1 object" and I don't get any further. Obviously I want to ungroup it to manipulate the image. It has colours, gradients and maybe transparencies.
First, please always attach or give a link to the file you refer to.
The URL is http://ian.umces.edu/discforum/index.php?topic=178.new;topicseen#new The file is ianfish.svg.
Erik
On 9/22/05, Erik <kaver@...68...> wrote:
The URL is http://ian.umces.edu/discforum/index.php?topic=178.new;topicseen#new The file is ianfish.svg.
Thanks, I tried it and found that my advice to unlink the clone wouldn't actually work for this file. That's because it uses a clone of <symbol> which we don't fully support. But I just checked a quick fix that will allow you to at least unlink the symbol properly. So with CVS version, you can ungroup, unlink, ungroup etc. down to the basic elements. There are still many glitches remaining (fish shifts on unlinking; dark polka dots disappear on ungrouping - but they're actually there, just save and reload the file to restore them; warnings in the console etc) but overall it works now.
-- bulia byak Inkscape. Draw Freely. http://www.inkscape.org
----- Original Message ----- From: "bulia byak" <buliabyak@...400...> To: "Erik" <kaver@...68...>
Second, judging by your description, that object is a clone. Press Shift+D to select to its original.
1. Select fish. Inkscape says "Group of 1 object" 2. Shift+D. Inkscape says "Select a clone to go to its original. Select a linked offset to go to its source." Then the message goes back to 1.
If it says the original is in defs
It does not say this.
and cannot be selected, then you can just unlink the clone by Shift+Alt+D.
3. Select fish. 4. Shift+Alt+D. Inkscape says "No clones to unlink in the selection". Then the message goes back to 1.
After that it's a regular object (in this case a group).
5. Nope. I still cannot ungroup it.
Erik
OK. I'll test CVS soon. In meantime I have a file which is ungrouped and has no wrapper from IAN and this is ungrouped when brought into Inkscape.
Now assuming that the import/ungrouping in Inkscape is fixed up one way or another my next question is "Has anyone managed to export a file from Inkscape in vector format that will open in CorelDraw? .ps is OK for paths and solid colours but does not work for gradients and these IAN files are full of gradients. I know I can get the full picture out as a .png file but I want to keep it as vectors. Anybody succeeded with this?
Erik
On Tue, 2005-09-20 at 17:35 -0400, Greg Steffensen wrote:
Ok, the clipartbrowser cvs module is now synched up again with the latest tarballs, and the cannonical download location for releases is now
http://www.python.org/pypi/clipartbrowser
(it used to be on Berlios). If you download a release (they work better than ever :), install using "python setup.py install" (assuming you have the dependencies met). The "easy_install" method mentioned earlier works too.
Cool...I still think you should do releases and use Inkscape's sf.net space just for the sake of affiliation and encouraging other developers who already have access to work with you.
Providing easy access to others is the key to getting more feedback and others hacking on your projects...and in order to take advantage of this, you should try to work in the community with the practices and traditions that the majority of developers know.
For example, I can work on your app when you use standard infrastructures, as I know standard Makefiles and packaging, but now am discouraged from working on clipartbrowser because I don't have time to learn "Yet Another Language."
How is the project going btw? I've been swamped in with new gigs -- teaching at sfai.edu and working at creativecommons.org
Jon
On 9/20/05, Greg Steffensen <greg.steffensen@...400...> wrote: I didn't explain the technical situtation with the install method very well. Basically, if I write a file "setup.py" that defines certain information, users should be able to execute "python setup.py install" to install the program correctly, and I should be able to execute commands "python setup.py bdist_win" to create actual win32 EXE files containing pretty automated installer GUIs. In addition, users will be able to install the way I described in the release email as well. In general, setup.py uses the "setuptools" which are the successor to the "distutils", which is the distribution/installation tools included with python. I'm still not certain whether I should use the setuptools or the older distutils, and am figuring stuff like that out, but they're both supposed to be the python-friendly replacement for Makefiles, and I'm trying to reorganize the code to do stuff the "correct" way.
Greg On 9/20/05, Greg Steffensen <greg.steffensen@...400...> wrote: Hey, yeah, this is kinda what I was referring when I said that the packaging isn't as polished as it needs to be. For what its worth, most of the changes (bug fixes, performance improvements) actually are committed to CVS; I just didn't commit commit the change in the install method simply because I was having trouble figuring out how to get files like COPYING, NEWS, etc. included using that method, and I didn't want to put a totally messed up version into CVS. Basically, I actually really released more on a whim, not because I'd worked everything out, but because I'd gone too long without releasing code, wanted people to know that work was ongoing. So, at a minimum, I should have indicated that this was definitely a development release. So, solutions... in terms of where formal releases (formal tarballs) should go, I actually think the python package index is definitely the correct place... its now the "official" place where the python folk reccomend that python projects be listed, and it will offer a number of technical advantages, including publication of new releases, and installation automation. I'm very happy keeping CVS at Inkscape though (not that PyPI offers CVS anyway), and treating the code as a subproject of Inkscape. But as long as it offers the ability to be installed standalone as well, I'd like to make standalone tarballs available from PyPI. In general, I'm learning more about python, and am trying to make my practices for installation, etc. match accepted practice. I'll try to get CVS synched up with the current state of the project very soon; I'm still having some trouble with the python installation tools, and have sent an email to their mailing list for advice, but haven't heard back yet. Once I know how to get the documentation files included, and where to put the main install file (setup.py), I'll get that committed and CVS will contain the official version of the code gain. Again, I would have waited until this was done to release, but just felt bad about having gone too long without releasing already, and was overeager to get something out. Anyway, does this cover the stuff you were concerned about? If not, let me know. Also, you may recall that someone posted a link to a mockup for a very similar project that they'd found, and I replied that this was "very depressing" :). I used that mockup to develop version 0.4, and emailed the designer (as the community suggested); he's finally written me back, and said he's interested in contributing, which rocks. Later, Greg On 9/19/05, Jon Phillips <jon@...235...> wrote: <offlist /> On Mon, 2005-09-19 at 05:20 -0400, Greg Steffensen wrote: > I'm releasing another version of the clip art browser. Lots of bug > fixes, some performance improvements, code cleanups, and a new > installation procedure (yes, yet another). > > Installation is now 3 steps, and hopefully easier than ever: > > 1) Satisfy the dependencies of Python 2.4 and PyGTK 2.6. > > 2) Download and run the following script, which installs the python > setuptools: > > http://peak.telecommunity.com/dist/ez_setup.py > > 3) Run "easy_install clipartbrowser" > > As you can see, I've switched the installation system from Makefiles > to the python setuptools. The setuptools are still under development > (they're apparently going to be included in python 2.5), and I'm still > very much learning how to use them, so the packaging isn't as polished > as it needs to be, but this makes installation a breeze. Eventually, > it should allow arbitrary python dependencies to be intelligently > installed as well using the same process. > > I'm still working on getting this integrated back into the Inkscape > effects menu, that's next on the list. As always, feedback greatly > appreciated. > Greg, I thought we were going to do the next release from Inkscape CVS? What you have done is a non-standard release of the project now that we consolidated the code into Inkscape. Also, you have now used a non-standard approach to packaging which then again breaks what users and developers are use to. The major thing though is that we were going to do a solid release which means you need to involve the people who would contribute to this in the release to help smooth out the bugs. Instead, now you have released from another location thus confusing users and developers more. The problem with this is that it is not pro-community. Anyhow, I'm curious what you think and also want to get these things on track with the community. It is so great that you are rocking these changes and I'm so proud of what you are doing so I don't want to dismay you, but really I think the next release has to be really on track with the community and we should coordinate it, push it out through the sourceforge system we have in place, and push the press release globally. Without these procedures, honestly, it is not very likely that others will help on development nor use your work. So, lets talk some more and sort these things out. Jon -- Jon Phillips San Francisco, CA USA PH 510.499.0894 jon@...235... http://www.rejon.org MSN, AIM, Yahoo Chat: kidproto Jabber Chat: rejon@...896... IRC: rejon@...897... Inkscape (http://inkscape.org) Open Clip Art Library (www.openclipart.org)
participants (5)
-
bulia byak -
Erik -
Greg Steffensen -
John Cliff -
Jon Phillips