CVS Checkout and Branches

[Ingo Mierswa]

Hi,

I just wanted to let you know that you of course still can checkout the latest version of RapidMiner via anonymous CVS like it is described here:

http://rapid-i.com/content/view/25/48/


There are several things to note:

• The branch "HEAD" is now a bugfix branch which is based on the latest official release and contains all bugfixes since then.
• New features will no longer be added to the HEAD branch but to the latest developer branch. Currently this branch is named "Wasat".
• There is a delay between our developer CVS servers and the anonymous CVS servers. It might happen that the synchronization takes several hours.

Cheers,
Ingo

[DrGary]

What's the relationship between the releases and the CVS repository? For example, I haven't found the new 4.5 operators anywhere, including head... What's the relationship between Yed, Wasat, Zaniah, and rincewind?

Thanks

[Sebastian Land]

Hi,
let's first explain the easy things: The names are the internal codenames for one version of rapid miner. So, Wasat is 4.5, Yed was 4.4 and Zaniah was 4.3. Rincewind, as immediately understandable for all who know this famous character of Pratchett's discworld, is a failed branch, so just ignore it If you take a look at the branches, you will recognize some sort of order and a common theme, at least if you are familiar with astronomy, and probably will be able to predict which name the next version will have

Currently we have some problems with the sourceforge cvs. In the past we were so frustrated from waiting half an hour for performing an update, that we moved to a local repository. Our admins are currently working on mirroring the local repository to the sourceforge cvs, so that one can checkout the latest, bugfixed version of each branch. Until they got this to work, you might use the downloadable source code, which is in fact the latest version until now.

Greetings,
  Sebastian

[kochan]

Hi,

Just to be perfectly clear about the cvs problems. Does it mean that the code that I check out from the sourceforge repository is not the latest version? And does that mean that there is no cvs access to the latest version from the outside?

I have checked out the Wasat branch, which I expected to be the 4.5 version, but if I look in the build.properties file it says 4.4.

Regards,

Andreas

[Sebastian Land]

Hi Andreas,
our admins have found a solution for that. BUT: Unfortunately we need a password from someone who is on vacation right now...
So until he returns, you will NOT have access to the latest source files from the repository.
The good news on that topic: When he returns we will switch to SVN on sourceforge, what should increase the performance.

Greetings,
  Sebastian

[phb]

i cant find any branch for version 4.6. any hints?

[Sebastian Land]

Hi,
as you might have read in this topic there is no recent version of RapidMiner in the sourceforge repository. Unfortunately sourceforge blocks a command, necessary for synchronizing our local repository with the sourceforge one. And they don't seem to be very eager to change that, also there is already a suggestion on their feature request site. If you have a sourceforge account, you might very well vote for this feature.
Until then our admin uses his free time (which is small enough as always)  to find some way around. Until then, or until sourceforge changes this one setting, you might use the version which comes with RapidMiner. The 4.6 version is up-to-date since we are mainly focusing on 5.0 now.

Greetings,
  Sebastian

[holger]

Hello Sebastian.

The good news on that topic: When he returns we will switch to SVN on sourceforge, what should increase the performance.

Any progress with migrating to svn? Being a sf.net project, hosting of the sources in a public code repository seems to be a more natural solution compared with a the currently used closed SCM. This would allow developers to contribute bug-fixes/patches and to provide conceptual feedback to make RM even better. :-)

So even if imho an unusual question for an "open source" project:
Is there any possibility to get a recent src-snapshot or RM5, which would allow to inspect the new operator/process-API? Maybe this could include even a draft of the (partially) updated documentation...

Best, Holger

[Sebastian Land]

Hi,
the current stable version is already mirrored on the SVN on source-forge. RapidMiner 5 will be there as soon as we publish the final version.
You might take the API from the Beta version anyway, there hasn't been conceptual changes, if I remember correctly.

Sorry for the inconvenience, but did you ever worked on sf with a big project like that? Everyone of us waited half an hour per day just for synchronizing. That was too disgusting. And in the past the number of contributors was not that big, that it was worth having these problems...

Greetings,
  Sebastian

[holger]

Hi Sebastian,

thanks for your quick response. To allow other RM-newbies to be less confused, you could consider to change http://rapid-i.com/content/view/24/47/lang,en/ as this is still pointing to cvs and not svn.

the current stable version is already mirrored on the SVN on source-forge. RapidMiner 5 will be there as soon as we publish the final version.
You might take the API from the Beta version anyway, there hasn't been conceptual changes, if I remember correctly.

According to what I read in the RM-forums, the plugin-API changed significantly between 4.X and 5. Do you refer to the API-doc in the sf.net-release? Or is there a source-snapshot available somewhere else for RM5b?

Sorry for the inconvenience, but did you ever worked on sf with a big project like that? Everyone of us waited half an hour per day just for synchronizing. That was too disgusting. And in the past the number of contributors was not that big, that it was worth having these problems...

I worked for years with the Sphinx-project on sf.net, which is an open-source speech recognition system. The HEAD was around 800mb (which is 10xRM), because we also included some test-speech corpora. And according to my experience, sf.net-svn-performance was quite alright during the last time. For sure we didn't do repository syncing but the usual 'svn ci/co'.

Concerning contributors: I clearly see your point, however, contributors and SCM (not just an occasionally mirrored one, which does only partially reflects how a projects evolves) are linked to each other. Personally, I would probably work out a patch for a particular problem, only when I'm sure that the problem has not yet been addressed. But without having (at least read) access to the development-SCM, this is something quite hard to figure out. For sure, I could write mails, file tracker issues and so on, but than even simple patches blow up to a considerable amount of work.

So if svn-performance is really an issue for you and sf.net-svn not an option, you could/should provide read-access to your rapid-i-internal SCM, which would fix both problems and would make everyone happy. :-)

Don't get my criticism wrong, I just want figure out to which extent I can rely on the openness of RM before pushing the people in my group away from Knime towards RM. :-)

Greetings, Holger

[Ingo Mierswa]

Ok guys,

please calm down. The source code of RM5beta was part of the release. Just checkout the "src" folder of the release and you will get the snapshot of the source code of the released version.

And of course you are right and giving access to the sources will ease patching etc. for the developers in our community. So this is high on our agenda. But as Sebastian has pointed out, we had some troubles with SF in the past and it least managed to get the stable branch (RM 4.6x) up again. And also the code access to the latest developer version of RM 5 is on it's way so please be patient enough for just a few days. Just think of it as another X-mas present 

I just want figure out to which extent I can rely on the openness of RM before pushing the people in my group away from Knime towards RM. :-)

The core of RM has always been and always will be open source. This was true since the very first days of its development and even in times where Knime had used this creepy only-partly-open-source-license-for-some-users-but-not-for-others 

So you can be assured that RM is as open as it can be...

Cheers,
Ingo

[holger]

Hi Ingo,

Just checkout the "src" folder of the release and you will get the snapshot of the source code of the released version.

That's the answer I needed.  Now I can enjoy to read the sources and wait for the release (getreu dem Motto "Vorfreude ist die schoenste Freude").

Just think of it as another X-mas present 

I love Christmas 

even in times where Knime had used this creepy only-partly-open-source-license-for-some-users-but-not-for-others

creepy: +1
It's great news for me, that you feel the same about the "open(fake)source"-approach of knime.

Best,
Holger

[Ingo Mierswa]

Hi Holger,

so have fun with your additional present 

Cheers,
Ingo

[Simon Fischer]

Hi all,

the 5.0 (Vega) branch is back on sourceforge now:

https://yale.svn.sourceforge.net/svnroot/yale/Vega/

Cheers,
Simon

[holger]

Great work! That's how opensource should look like.  Thanks a lot. It compiles and runs out of the box. :-)

-holger