Please use this thread to post any comments/questions you have about Intel Mash Maker.
I was able to immediately login, but after recycling FF 3x and even rebooting, I still have not received the downloaded mashup database. I checked my Firewall and it doesn't seem to be blocking anything from Intel, so I just wanted to check here and see if it was just my end. TIA. Damn this is gonna be cool!
Am investigating this at my end right now.
Seems we received your request to download the database, and I don't see any record of an obvious error that would have caused us to not send it to you.
I'll try adding in some more debug code and see if that sheds any light on it...
It's possible the firewall might be the issue - Mash Maker hasn't been tested with a firewall yet...
Is anyone else having this problem?
Still can't find anything wrong. I can log in with your account and it seems to work fine.
You might want to reinstall the plugin with the version up on the server now, just to ensure you are running exactly the same version as I am.
What OS are you running on? Are you running the latest version of firefox? Do you have any other plugins installed?
Good morning Robert, thanks for helping out.
I just added a rule to my Comodo firewall allowing the http://mashmaker.intel.com IP IN or OUT and then rebooted. I still get the msg on my Mash Maker toolbar saying it can not dl the database though.
I'm on an up-to-date XP box, with current FF and too many ext.s to mention: http://www.billy-girlardo.com/My_Firefox_Config.html
I suspect this will be a problem for others as well...
I'm a big fan of Intel's previous releases of free/libre software, such as PowerTOP (http://www.linuxpowertop.org/) and various Linux kernel patches. It seems that the Intel Mash Maker client and server software would benefit from being released as free/libre software as well. It would drive adoption and allow programmers to help out (I think there are many who would be interested in this).
By releasing the server code, people with spare server time could start running their own Mash Maker servers, allowing more people to test the code while Intel has fewer server resources allocated during the testing period. Instead of handing out the software to select people, Intel could hand out the software to everyone, but handout logins to Intel servers to select people, while others can use community-run servers (ie. servers not run by Intel).
Is Intel planning on releasing the Intel Mash Maker client and server code under a free/libre software license?
This problem should now be fixed. If anyone else was having this problem, please go back to the download page and re-install the plugin.
Releasing this as open source is definitely an option, and is something that appeals to me personally.
No commitment has been made either way about how we want to license Mash Maker long term. The license on the preview release is just a default license for technology preview releases.
Watch this space...
The error msg went away, thanks again.
I had it working just fine earlier today. I had to restart Firefox to install Firebug, and now the database won't load.
I'm getting a 503 on the download page. Is your server down?
Yep. It's a server problem. Tomcat keeps crashing for some reason. We are currently rewriting the server front end as PHP, which will hopefully make things more stable...
I've been getting server connection errors since yesterday. I restarted firefox and now I'm getting the same old "Could not download mashup database" error.
I tried to download the add-in and re-install, but got another 503. Looks like your server is down again.
As of about a week ago we've been using the new PHP based server. Hopefully nobody should be seeing any of those "server not available" or "cannot download mashup database" errors anymore.
Let me know if you have any problems.
I like having a IRC channel to pop into when I'm struggling with small questions that don't warrant a forum post. Unfortunately, I have zero experience managing a IRC channel, so I can't really make one myself. How about claiming #intel-mashmaker on irc.freenode.net?
Some questions/problems I ran into:
How do you expand relative urls from href attributes into full ones? Most mashups seem to only work if objects have full url properties. Can you prefix the XPath result with your own string as a workaround? Or maybe a new type 'relative url' that gets recognized and reformed into a full url by the mashups is in order here.
It isn't clear what extractor the extractor edit side-bar is showing for a webpage. If there are more then 1 applicable extractors for an url, it still only shows one, and you seem to have to switch between all the possible extractors by using the 'History...' button. Even then, it shows extractors in the History overview that do not match the url of the current page. How does it know what extractors to show in the History overview?
When editing an extractor for a page and switching tabs it asks you whether you want to clear it. The effects of clicking 'cancel' or 'ok' here are not totally clear to me. It seems 'ok' means 'revert this extractor to the last local version' and 'cancel' means 'save changes to the local version'. Closing a tab and later returning to that site and opening the extractor edit bar magically opens your saved local copy.
Versioning systems can be a tricky UI challenge. The current way is workable (and UI improvements can easily be procrastinated), but I would like to see an extractor browser that also shows the following:
EDIT: Oh yeah. I tried out the "Edit this mashup..." functionality and inadvertently published a new version with no changes because I didn't think the "Ok" button would publish it. That window really needs a "Cancel" button, and the "Ok" button should be called "Publish".
EDIT2: When sorting columns in the 'table' mashup it looks at the type to know if it should order in lexicographical or number order. However, the type selector overview does not show you which ordering a type adheres to. This ordering property could be expressed in some generic supertypes that at the same time give hints to parse the character data (for stripping currency signs, dealing with floating point numbers and scientific notation).
EDIT3: The annotate mashup clashes with XPath expressions looking for 'tr' elements. It finds the tr element used to display the annotation, and leaves an extra empty tr element in the web page each time you enable and disable the mashup. For an example, try enabling and disabling the annotation mashup on http://forums.penny-arcade.com/forumdisplay.php?f=20 a few times.I also have problems with the annotate mashup understanding the url properties to put itself under, but this might be related to the problem
above. You can also see this on the aforementioned page.
I love it so far though, keep up the good work!
How to create a new extractor? I don't have a menu item named "Data Tree Sidebar".