Kepler: Issueshttps://projects.ecoinformatics.org/ecoinfo/https://projects.ecoinformatics.org/ecoinfo/ecoinfo/favicon.ico?14691340362008-04-16T23:48:32ZEcoinformatics Redmine
Redmine Bug #3231 (New): 3-way table output (multiway cross tabulation actor?)https://projects.ecoinformatics.org/ecoinfo/issues/32312008-04-16T23:48:32ZKevin Drurydrury@nceas.ucsb.edu
<p>3-way tables are a useful extension of 2-way tables, which are a basic tool for exploring relationships in data. R has devised a simple way of displaying 3-way tables in the console. I don't know what the Multiway Cross Tabulation (MCT) actor is supposed to do (it is under development?), but it falls under the Statistics category of Kepler components and the name got me thinking it is intended for this purpose. I dunno. But, if something like ftable() in R is the intention of that actor, it would definitely be a useful addition to statistical analyses in Kepler. If that is not the intention of the MCT actor, may I suggest an actor that does this. Specifically, from the R Help:</p>
<p>Displaying a contingency table in this flat matrix form (via print.ftable, the print method for objects of class "ftable") is often preferable to showing it as a higher-dimensional array.</p>
<p>Thanks!</p> Bug #3230 (New): Matrix Viewer actor does not accept RExpression matriceshttps://projects.ecoinformatics.org/ecoinfo/issues/32302008-04-16T23:40:04ZKevin Drurydrury@nceas.ucsb.edu
<p>2-way tables are a common way to view relationships between bivariate data. The The RExpression actor is quite good at generating these and the Matrix Viewer actor seems like a logical vehicle for displaying them. The Viewer does not display the matrix, however, probably because, as the documentation says, it "...accepts matrix tokens (e.g., [1, 2; 3, 4]...", which is not the way matrices are represented in R (though they can be in Matlab). I am way out of my depth in discussions of the way such data is stored in, and output from, the RExpression actor (or in R for that matter). But, maybe I can say that if you want folks to be able to generate high-quality output in workflows, adequate presentation of a 2-way table would be a pretty basic expectation.</p>
<p>Thank you!</p> Bug #3226 (New): Eval: subplots from RExpression overly condensedhttps://projects.ecoinformatics.org/ecoinfo/issues/32262008-04-16T22:42:19ZKevin Drurydrury@nceas.ucsb.edu
<p>Using nightly build 14 Apr 08 under WinXP. The code below generates a publication-quality graphic when run from the R command line. When run in an RExpression actor, however the results are pretty unattractive.</p>
<p>Case 1: With an imageJ actor connected, the resulting RExpressionNUM.png is extremely compressed. i.e., there is lots of white space and the subplots are cramped.</p>
<p>Case 2: With no imageJ actor, and the "Automatically display graphics" button checked, the resulting pdf is less cramped than the png, but still not as well-proportioned as that coming directly from R.</p>
<p>Note: resizing the graphic does not help - proportions remain the same</p>
<ol>
<li>test code for RExpression actor subplot generation</li>
<li>get data and variables in order...<br />library(MASS) ;<br />names(wtloss) ; attach(wtloss) ;<br />w<-Weight ; d<-Days ;</li>
<li>set up stacked subplotting environment, plot top fig<br />par(mfrow=c(2,1)) ; plot(w~d) ; <br />res<-lm(w~d) ; abline(res) ;</li>
<li>plot bottom fig<br />r<-residuals(res) ; plot(d,r) ;</li>
<li>return to default graphics, clean up<br />par(mfrow=c(1,1)) ;<br />detach(wtloss) ;</li>
</ol> Bug #3223 (New): interactive graphics in Rhttps://projects.ecoinformatics.org/ecoinfo/issues/32232008-04-15T21:34:15ZKevin Drurydrury@nceas.ucsb.edu
<p>Hello,</p>
<p>Just fyi, the identify() function does not seem to work when called from the RExpression actor. (I'm using Windows XP, the 14 Apr nightly build of Kepler, R 2.6.2)</p>
<p>When called from the R command line, this function waits for a pre-specified number of mouse clicks, then inserts labels at those (x,y)-coordinates. It's handy for labeling points, but not essential. I only bring it up because the reason(s) that it doesn't work via the RExpression actor may be related to other issues in past or future, etc.</p> Bug #3222 (New): save new workflow templatehttps://projects.ecoinformatics.org/ecoinfo/issues/32222008-04-15T21:07:19ZKevin Drurydrury@nceas.ucsb.edu
<p>Hello,</p>
<p>This is a suggestion about usability, not a bug. It would be very nice to be able to save a workflow that contains the basic components of a "typical" analysis. For example, I find that I almost always need an SDF director, an RExpression actor, an ImageJ actor, and a text display actor. I noticed on the File menu that the "New Workflow" has a submenu. How hard would it be to allow users to save a typical analysis design as a template and add it to this submenu? (in the meantime, I'll make a "template.xml" with these components, open it, save as "newfile.xml" and keep on truckin')</p>
<p>Thank you!</p> Bug #3192 (New): behavior of documentation actorhttps://projects.ecoinformatics.org/ecoinfo/issues/31922008-03-31T23:38:46ZKevin Drurydrury@nceas.ucsb.edu
<p>Hello. The documentation actor will be an important component of workflows that outlive individual analyst's involvement in a project. Although the canvas provides an intuitive picture of the analysis, it doesn't take much complexity before some careful documentation is worthwhile. To some extent, annotations can do the trick, but I have found that sometimes we want a few paragraphs about the intended use, followed by a list of steps to follow. This seems especially important when the workflow serves an exploratory function and various actors get connected and disconnected as the work proceeds. I finally landed on the Documentation actor, but someone may know of a better solution?</p>
<p>Anyway, in dealing with the Documentation actor, I found different behaviors on the mac and pc. Specifically, I was able to get documentation for my workflow on a MacBook Pro, and edit it, but after opening the same workflow on my pc, I could only see the documentation once. On the mac, I dragged a Documentation actor to the canvas, right-clicked the canvas and chose Documentation->Customize or Documentation->Display. On the pc, the Display option works once, but then Customize and subsequent requests to Display show empty documentation forms, one for customizing, one for displaying - but no documentation exists. Is it me???</p>
<p>Also, it would be very helpful if workflow documentation would retain formatting. For example, on the mac where I've had the most luck creating documentation, I format with paragraph breaks and numbered lists - all the usual ways of breaking up text into logical and useful groupings. And, this formatting is retained in the Customize window. It displays as one long string in the Display window, however. Is this easy to fix?</p>
<p>Thank you!</p> Bug #3187 (In Progress): RExpression actor does not share focushttps://projects.ecoinformatics.org/ecoinfo/issues/31872008-03-26T00:29:57ZKevin Drurydrury@nceas.ucsb.edu
<p>It would be nice to be able to change the focus from an open RExpression actor to another Kepler window. Two cases that commonly occur:<br />1) copying code from one RExpression actor to another - when the source code comes from several regions of an RExpression R-script this leads to copying lots of stuff you don't need (that is in between the parts you do need), or opening and closing the 2 RExpression actors multiple times. Because of the tiny size of the R-code window, believe it or not, it's the latter path that I usually choose. Especially before this window was scrollable - whoever fixed that, thanks!<br />2) often, the next change an analyst will make to their code depends on what the output looks like from the last execution. So, you look here in the output window (say, a standard text output actor), then you scroll down and look there, and other places in a long output stream. Then, you open your RExpression actor to make the desired changes, but suppose you need to double-check a bit of output that is no longer visible, because you've scrolled past it. You have to close the RExpression actor in order to scroll back through the output to the spot you need. Then reopen the actor. But, now other parts of the output are invisible and don't become visible (by scrolling to them) until you close the RExpression actor again.</p>
<p>Is it possible to allow back-and-forth between 2 open RExpression actors and an RExpression actor and the workflow output?</p> Bug #3172 (New): windows mouse wheel doesn't scroll canvashttps://projects.ecoinformatics.org/ecoinfo/issues/31722008-03-11T23:04:12ZKevin Drurydrury@nceas.ucsb.edu
<p>Kepler Version: Release Candidate 1<br />OS: Windows xp</p>
<p>I have a usb mouse. When a workflow exceeds the window size, vertical and horizontal scroll bars appear. It would be nice to be able to scroll the canvas using the wheel of the mouse. Is it easy to do? I believe it is the common user's basic expectation. While I'm on the subject, it would also make sense for the keyboard arrows to scroll the canvas.</p>