Rev 475 | Blame | Compare with Previous | Last modification | View Log | Download | RSS feed
Note that all further changes will be noted in a separate file(SvnLog).2004-04-08 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/qq.R, R/qqmath.R, man/qq.Rd, man/qqmath.Rd: changed defaultbehaviour of qq and qqmath to use actual data instead of samplequantiles. The new behaviour is different from the correspondingS-PLUS functions, but is closer to analogous base functions qqplotand qqnorm2004-03-22 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/print.trellis.R: changed behaviour of skip* man/xyplot.Rd: updated docs for skip2004-03-19 Deepayan Sarkar <deepayan@stat.wisc.edu>* utilities.3d.Rd: removed palette.shade from docs* panel.cloud.Rd: adjusted docs for panel.3dwire's shade.colors arg* latticeParseFormula.Rd: fixed docs for drop* R/*.R: added drop.unused.levels as a high-level argument toxyplot etc. The earlier behaviour was a trellis.par.get-tablesetting, but that doesn;t make sense because this is not a displayparameter2004-03-17 Deepayan Sarkar <deepayan@stat.wisc.edu>* panel.functions.Rd: added panel.rug in docs* lset.Rd: added link to trellis.device2004-03-13 Deepayan Sarkar <deepayan@stat.wisc.edu>* panel.cloud.Rd: added note saying x,y,z could all be matrices* cloud.Rd: added docs in wireframe for parametrized 3-D surfaces2004-03-12 Deepayan Sarkar <deepayan@stat.wisc.edu>* DESCRIPTION: updated version* tests/test.R: minor changes* demo/lattice.R: added wireframe sphere* R/densityplot.R: fixed default prepanel function to not throw anerror (in density()) when panel has only 1 point* R/common.R: modified latticeParseFormula to keep matricesunchanged (so that a formula of the form matrix ~ matrix * matrixcan be used in wireframe)* R/cloud.R: added (way cool) changes to allow wireframe to render3-D surfaces parametrized on a grid (e.g. a sphere)* man/update.trellis.Rd: added mention of legend* man/panel.pairs.Rd: fixed synopsis* man/print.trellis.Rd: fixed grid links (hopefully)2004-03-08 Deepayan Sarkar <deepayan@stat.wisc.edu>* miscellaneous.R: added a col.line arg for lplot.xy todistinguish between color for lines and points (not that itmatters here, but panel.xyplot and panel.superpose make thedistinction, and it's easier to just pass it down here)* cloud.R, levelplot.R, scales.R, settings.R, update.trellis.R,zzz.R: many improvements to cloud and wireframe, plus misc changes2004-03-05 Deepayan Sarkar <deepayan@stat.wisc.edu>* tmd.R: tmd simplified using update(), fixed potential xlab/ylabbug* miscellaneous.R, print.trellis.R, update.trellis.R:obscure improvements to axis setting usage and update method* R: bunch of updates, mostly dealing with fonts (fontfaces,fontwidths) and labels (main, sub, xlab, ylab)2004-03-03 Deepayan Sarkar <deepayan@stat.wisc.edu>* man/xyplot.Rd: added docs for perm.cond, index.cond; updatedother parts* R: bunch of fairly major changes, including code reorganizationand addition of features, too numerous to list. A summary of thechanges is available in inst/Changes2004-03-02 Deepayan Sarkar <deepayan@stat.wisc.edu>* man: doc updates mostly to ensure R CMD check passage2004-02-14 Deepayan Sarkar <deepayan@stat.wisc.edu>* xyplot.Rd: added entry for par.settings, corrected statementsabout subscripts2004-02-13 Deepayan Sarkar <deepayan@stat.wisc.edu>* xyplot.Rd:added note on panel.number and panel.counter arguments. Has a relatedbrief and oblique mention of the conditioning variable permutation /reordering stuff. Not sure what the best place to put a more detaileddescription is.2004-02-10 Deepayan Sarkar <deepayan@stat.wisc.edu>* man/Lattice.Rd: wireframe is no longer slow, so removedstatement to that effect. Replaced S-Plus with S-PLUS2004-02-08 Deepayan Sarkar <deepayan@stat.wisc.edu>* man/print.trellis.Rd: added warning not to use aspect != "fill"with explicit panel.width and panel.height2004-02-05 Deepayan Sarkar <deepayan@stat.wisc.edu>* man/xyplot.Rd, cloud.Rd: fixed errors of the form\link{\code{stuff}}2003-12-20 Deepayan Sarkar <deepayan@stat.wisc.edu>* DESCRIPTION, NAMESPACE : made changes to deal with packagereorganization in R 1.9.0-pre. Some other files changed as well(mostly dealing with loess.smooth, which is now in stats)2003-10-17 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/axis.R : added trim = TRUE in format() when calculating axistick labels. Without this, labels often had leading spaces,causing a visible 'right shift' (reported by Patrick Connolly).2003-10-10 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/ : split common.R into common.R, axis.R and scales.R2003-10-09 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/levelplot.R: Fixed panel.levelplot bug where nothing was drawnif a panel contained only one unique value of x/y (reported byWolfram Fischer).2003-09-15 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/miscellaneous.R: added some code to handle cex when pch =character, using new grid feature. However, grid has a bug whenpch - numeric, so there's still a workaround in place, whichshould eventually be removed. This overrides related changes on2003-09-03* man/panel.pairs.Rd: added doc for prepanel.limits, pscales* tests/test.R: added code with splom pscales weirdness* R/splom.R:earlier interpretation of pscales was ambiguous. Changed to makespecification of limits, at and labels consistent. See panel.pairsdocs.* man/splom.Rd:moved bulk of pscales doc over to help page for panel.pairs2003-09-05 Deepayan Sarkar <deepayan@stat.wisc.edu>* splom.R:added a prepanel.limits argument to panel.pairs to allow custom axisrange calculations for each variable2003-09-03 Deepayan Sarkar <deepayan@stat.wisc.edu>* miscellaneous.R:lplot.xy modified to use grid.text when pch is non-numeric. Thiswould allow pch like pch = 'string' or pch = expression(sigma). Thisfeature would disappear in the future.2003-08-29 Deepayan Sarkar <deepayan@stat.wisc.edu>* levelplot.R:the older contourplot routine went horribly wrong when there were NA'sin the z-values. This has now been addressed with a slightly crudehack: all z-NA's are now replaced by min(z, na.rm = TRUE) for thecontour drawing routine. This may cause inaccuracies near the"border", but should be OK in most real-life cases, where the non-NApart should be a largish connected piece, with a fairly insignificantboundary.This incidentally fixes another bug, apparent when the at vectordoes not cover the full z range (colors used to be recycledinstead of leaving blank spaces)2003-08-22 Deepayan Sarkar <deepayan@stat.wisc.edu>* settings.R: (canonical.theme) also added clip option for the strip* print.trellis.R:clipping for panel and strip now determined by trellis setting "clip"* settings.R:added a clip = list(panel = TRUE) to the list of valid settings2003-08-21 Deepayan Sarkar <deepayan@stat.wisc.edu>* levelplot.R (panel.levelplot):shrinking had problems for constant z values withina panel (diff(range()) = 0, leading to division by 0). Also, it turnsout that shrinking was being done based on range of z within eachpanel, not the overall data frame. This could be misleading, sochanged.2003-08-20 Deepayan <deepayan@stat.wisc.edu>* common.R (calculateAxisComponents): the format argument renamedto format.posixt (to avoid confusion with the formatfunction). More importantly, in the "plain numeric" case,as.character(at) replaced with format(at). So now at = c(1, 1.5,2) will be labeled as c("1.0", "1.5", "2.0") instead of c("1","1.5", "2"). Both changes suggested by Tom Short.* common.R, settings.R, zzz.R: Supplemental changes related toNAMESPACE. Also renamed man/lattice-internal.Rd toman/lattice-internal.Rd.foo --- this is ignored by R CMD scripts,but has been retained as a possibly useful reference for internal(unexported) functions.2003-08-05 Deepayan <deepayan@stat.wisc.edu>* zzz.R: NAMESPACE changes (due to Prof Ripley)* print.trellis.R:changes associated with NAMESPACE (actually moving .lattice.print.moreto an environment)* NAMESPACE: added NAMESPACE file* R/zzz.R, R/print.trellis.R, R/settings.R: changes related toNAMESPACES (actually due to moving a couple of global variables toa new environment (see R/zzz.R) ), due to Prof Ripley. Alsosimplified trellis.par.get() a bit.2003-05-31 Deepayan Sarkar <deepayan@stat.wisc.edu>* bwplot.R, cloud.R, miscellaneous.R, xyplot.R:changed order to sort.list where appropriate2003-05-29 Deepayan Sarkar <deepayan@stat.wisc.edu>* levelplot.R: added na.rm = TRUE to min in min(abs(slopes))because sometimes slopes can be NaN. Also, added graphicsparameters for panel.levelplot labels2003-05-24 Deepayan Sarkar <deepayan@stat.wisc.edu>* panels.R:added a function called panel.curve, similar to curve() in base.* R/panel.functions.Rd: added docs for panel.curve* man/panel.pairs.Rd: added doc entry for 'panel.number'2003-05-22 Deepayan Sarkar <deepayan@stat.wisc.edu>* tests/scales.R: added output file name* tests/MASSch04.R: changes a call to title() to main=* tests/test.R: added output file* tests/dotplotscoping.R: added output filename2003-05-22 Deepayan Sarkar <deepayan@stat.wisc.edu>* DESCRIPTION: updated DESCRIPTION* man/panel.pairs.Rd: added doc entry for 'panel.number'* R/common.R (limits.and.aspect): Following up on a report byKevin Wright, I discovered that the presence of blank panels(either by skip or due to layout specifying more panels than wereactually there) caused problems in calculating factor levels(adding an extra NA level). This should be fixed now. Also,handling of xlim=list(...) improved in cases with skip (not thatanyone is ever going to use it).2003-05-22 Deepayan Sarkar <deepayan@stat.wisc.edu>* splom.R:panel functions can now optionally ask for a panel.number argument,which would be supplied as an argument only if the panel function hassuch an argument. Unfortunately, this was not working as intended insplom, because of the unusual nature of the panel/superpanelarguments. This has been fixed.2003-05-19 Deepayan Sarkar <deepayan@stat.wisc.edu>* man: fixed minor doc problems to pass R CMD check2003-05-18 Deepayan Sarkar <deepayan@stat.wisc.edu>* R: Almost all files updated to fix bugs related to the scalesargument --- limits inside scales had a bad bug which is nowfixed, also added and improved support for multiple (as lists)limits, at, labels. axs = "i" should also work.* tests: new tests added* R/miscellaneous.R: modified lsegments to accept x0, x1 as wellas x1, x2 (a la S-PLUS).2003-04-10 Deepayan Sarkar <deepayan@stat.wisc.edu>* man: Documentation updates2003-03-26 Deepayan Sarkar <deepayan@stat.wisc.edu>* bwplot.R, common.R, panels.R, print.trellis.R, zzz.R:some bugfixes, added MASS ch04 script to tests2003-03-26 Deepayan Sarkar <deepayan@stat.wisc.edu>* print.trelis.R: fixed problem with expressions as axis ticklabels when relation = "same". Reverted back to using (slower) oldcode in colorkey because of (grid?) bug in new version.* common.R: added code in latticeParseFormula to convert matricesto data frames (for splom and parallel) before subsetting.2003-03-15 Deepayan Sarkar <deepayan@stat.wisc.edu>* common.R:fixed some problems when outer and conditioning variables interact* print.trellis.R: fixed y-axis scales drawing* common.R (latticeParseFormula):added argument outer=FALSE. When outer = TRUE,variables separated by + (when multiple = TRUE) are made into aconditioning variable instead of grouping variable* several: updated support for multiple LHS/RHS in formula,auto.key, outer2003-03-14 Deepayan Sarkar <deepayan@stat.wisc.edu>* levelplot.R: changed prepanel functions to handle factorsgracefully. Fixed problem with unequally spaced gridpoints. Improved contourplot labelling. New argument shrink inpanel.levelplot, allowing rectangle dimensions to be proportionalto height. (Many of these were actually done much earlier, butcommitted now)* bwplot.R:fixed some problems with processing of formula in dotplot etc beforecalling bwplot* common.R: removed latticeAsFormula, no use2003-03-14 Deepayan Sarkar <deepayan@stat.wisc.edu>* print.trellis.R (print.trellis):Among other things, a rewrite of how axis ticks andannotation are drawn (mainly how the parameters translate to physicaldimensions). The tck, rot and cex components of scales can now belength 2 vectors, for bottom/left and top/right scales. tck can benegative, giving ticks inside panels. panel functions with a... argument are now passed a variable called panel.number, giving thesequence number of the panel (not sure what happens when skip isnon-FALSE)* plot.shingle.R:modified plot.shingel to make tck, cex and rot 2-vectors2003-03-13 Deepayan Sarkar <deepayan@stat.wisc.edu>* bwplot.R: added code to panel.barchart (also panel.dotplot) to handlegroups. Substantially changed the way arguments are processed inbarchart, dotplot and stripplot before being passed on to bwplot. Wheny is missing, and x has non-null names(), names are used as the yvariable.Also added changes necessary to enable multiple LHS and RHS is formulawith allow.multiple = TRUE.2003-03-13 Deepayan Sarkar <deepayan@stat.wisc.edu>* common.R (construct.scales):made changes so that cex, tck and rot can now belength 2 vectors in scales, affecting scales at the bottom/left andtop/right of panels respectively. Associated changes in print.trellislater.2003-03-13 Deepayan Sarkar <deepayan@stat.wisc.edu>* settings.R (trellis.device):Changed the default for the theme argument from NULLto getOption("lattice.theme"), which can be either a list, or a(possibly quoted) function, which when called will return a list.2003-03-09 Deepayan Sarkar <deepayan@stat.wisc.edu>* splom.R (panel.pairs):the pscales argument didn't work when it was specifiedas a list. Fixed.2003-03-07 Saikat DebRoy <saikat@stat.wisc.edu>* common.R (latticeParseFormula):Compute a new element of the answer, subscr, if subscripts=TRUE.2003-03-07 Deepayan Sarkar <deepayan@stat.wisc.edu>* common.R (latticeParseFormula):added a subscripts = FALSE argument, which isnot used anywhere yet.2003-03-05 Deepayan Sarkar <deepayan@stat.wisc.edu>* print.trellis.R (print.trellis): added two new arguments toprint.trellis;panel.height = list(1, "null"),panel.width = list(1, "null"),which can be used to specify explicitly the dimensions of eachpanel in terms of grid units. Any unit is allowed as long as itdoes not need a 'data'. Not too difficult to add that, butprobably unnecessary. Behaviour is undefined (well, it is defined,but I don't want anyone to count on it) when aspect is not "free"in the object being printed. Docs not updated yet.2003-03-05 Deepayan Sarkar <deepayan@stat.wisc.edu>* print.trellis.R (draw.colorkey): Colorkey changed to use thevector fill gpar when drawing rectangles. This means allrectangles are drawn at one go, hence less things to grid.pack(),hence faster when lots of cuts in levelplot/wireframe.2003-03-04 Saikat DebRoy <saikat@stat.wisc.edu>* common.R (latticeParseFormula):Fix concat() to handle shingles correctly.2003-03-03 Saikat DebRoy <saikat@stat.wisc.edu>* R/common.R (latticeParseFormula):Allow the possibility of having multiplevariables in LHS or RHS of the model formula (eg. y1+y2 ~ x1+x2).2003-03-03 Saikat DebRoy <saikat@stat.wisc.edu>* R/common.R (latticeParseFormula):Use the environment of the model as enclosargument in calls to eval().2003-02-24 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/zzz.R: Restructured .First.lib following suggestions fromProf Ripley. Changed library(grid) to require(grid), with anerror message if that fails.* R/levelplot.R: xlab, ylab, etc in contourplot was not passed onproperly to levelplot. Fixed.2003-01-30 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/densityplot.R: changed prepanel.default.densityplot to returnrange instead of some-stupid-thing else when a groups argument isgiven.2003-01-16 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/settings.R: changed trellis.par.get so that when called withno device open, it starts the default device a la print.trellis2003-01-15 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/levelplot.R: levelplot used to calculate wx and wy, which wasnot a good idea since this should have been done inside the paneland prepanel functions. The prepanel function has been rewritten,fixing a bug, but some more reorganization still needs to be done.2002-12-18 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/miscellaneous.R: fixed problem the with 'adj' argument inltext (used to ignore second component, PR#2380). added 'pos'argument as in text. No offset yet, though.* R/panels.R: Changed panel.superpose to pass groups andsubscripts to panel.groups (PR#2377). Not very obvious how groupsmight help, but I guess more information is better thanless. Also, several other panel functions didn't recognize thecol.line argument, this was fixed.2002-12-15 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/common.R: The changes in lattice 0.5-0.6 added some code thatgave bad results when some panels had all NA's. That hopefullyfixed.* R/miscellaneous.R: lplot.xy used to call grid.text instead ofgrid.points when pch was a character. Fixed this.* R/print.trellis.R: draw.key failed when lines component hadtype='b', because of a missing pch. Fixed this.* R/levelplot.R: Minor fix related to changes in grid.2002-12-06 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/settings.R: added warnings and error messages intrellis.par.get/set when no active device, and when value intrellis.par.set is not a list.* R/trellis.par.get.Rd: rewrote some things2002-11-16 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/common.R: added arguments have.log and logbase tocalculateAxisComponents()* R/print.trellis.R: made changes to use above changes2002-10-28 Deepayan Sarkar <deepayan@stat.wisc.edu>* R: Hmm, lots of changes made without any log entries. Should bemore careful ! Anyway, renamed inst/README to inst/Changes, andadded a brief overview of changes in lattice 0.6 . Also made somedocumentation updates. Changed version to 0.6-6, ready to upload.2002-01-10 Deepayan Sarkar <deepayan@stat.wisc.edu>* R: Too many changes to list, mainly in print.trellis anddocumentation.2001-08-08 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/levelplot.R: Implemented preliminary levelplot (no contourplotthough). Also a version of cloud, though no support yet for niceaxes/arrows and stuff (does allow panel.superpose though). Again,no plans of implementing wireframe in the near future.2001-08-08 Saikat DebRoy <saikat@stat.wisc.edu>* R/common.R:Added support for 3d formulae in latticeParseFormula2001-07-26 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/Lattice.Rd: Included llines, lpoints etc for easy adaptationof existing Trellis code. qq() changed to as it should be, shouldnow work with tmd().2001-07-12 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/Lattice.Rd: Major redesigning of existing functions. Mainlyby collecting common code into a few functions and hencemaking individual functions easier to handle. Code should nowbe fairly readable. Also, more or less good support for log scalesand key (I hope). Not much checking done yet, there may be bugs.Documentation might be out of step at places.2001-05-06 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/tmd.R: Name changed from trellis to Lattice. Prelilminarychanges to account for change of Lattice to Grid made. Newfunctions rfs, oneway, qq (not final), qqmath, tmd, plot.shingle,update.trellis. Documentation more or less clean.2001-04-19 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/trellis.R: lots of changes. Introduced trellis objects.Print method for trellis objects. Implementation now includesbarchart, bwplot, densityplot, dtplot, histogram, stripplot,xyplot, splom and parallel (and excludes piechart(will probablyremain that way), qqmath, qq, timeplot(?), contourplot,levelplot, wireframe and cloud). Cleanup, documentation etc.2001-03-22 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/trellis.R: added splom and some brief documentation for splom2001-03-02 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/trellis.R: dtplot (dotplot) implemented. Probably has lots of bugs.Some documentation, including example for dtplot (same as usual dotplot example).2001-03-02 Deepayan Sarkar <deepayan@stat.wisc.edu>* R/trellis.R: updated bwplot (with the new layout structure)