MacClade 3 Bug List

Wayne P. Maddison and David R. Maddison
14 August 1996

Here is a list of the known bugs in MacClade 3.

Unless otherwise stated, bugs fixed in version 3.0x of MacClade were present in version 3.0 through 3.0x-1. (For example, most bugs fixed in version 3.03 were present in versions 3.0-3.02.) Unless otherwise stated, the bug was found by us and has not been reported by a user. Bugs marked with may affect the numerical results presented by MacClade.

Bugs fixed in 3.01

• For molecular sequence data, if States chart on screen, and data file saved, then first line in textual summary of chart will be bogus until you ask to have the chart recalculated. Bug in version 3.0.
• If you copy to the clipboard entire rows and columns from the data matrix, then select entire whole rows and columns that contain different numbers of taxa and different numbers of characters than those in the clipboard, and try to paste, MacClade will rightly refuse to paste the data, but it will incorrectly add new rows or columns or delete old rows and columns in the process. Bug in version 3.0.
• For protein data matrices, if you are in the treewindow, and ask for Type Edit under Assume, then press the New button, and choose "Protpars type based on genetic code", and press Create, MacClade will create a protein parsimony user-defined type. However, there may be amino acids in the matrix that are not present in the current genetic code (e.g., an amino acid specified as serine (S), but not specified as to which type of serine (1 or 2) in the standard nuclear code). MacClade would then present you with the query "AAs not present in genetic code found in matrix: do you wish to convert these to the appropriate AAs?" If you choose to convert, then various statistics, including traced characters, various charts, and so on, will not be updated. To have them recalculated properly, return to the data editor, and then go back to the tree window.
• If you ask to Recode a character in the editor, and Footstates is turned on, then the footstates will not be updated after the recode.
• Some of the functions for adding taxa and characters in the editor allowed you to expand the matrix beyond the allowed 1500 taxa and 16000 characters.
• If you created new user-defined types in the Type Edit dialog box, but pressed Cancel to exit the dialog box, MacClade will remove the newly-created types, as expected. However, the next time you choose the Type Edit dialog box, MacClade may crash. This crash can be avoided if you trace a character in the tree window before you call Type Edit the second time, or if you select one entire character in the data editor just before calling Type Edit.

Bugs fixed in 3.02

• Taxon, character, and state names cannot consist entirely of numbers. MacClade mistakenly allowed names that consisted of number followed by blanks or underscores. Bug reported by one user.
• If you copy a complex picture (perhaps a color image) to the Clipboard in another program, then switch to MacClade, MacClade may crash if that picture takes more than 32K of memory.
• There were a few minor problems with the vertical positioning of the tree in the treewindow and the vertical scroll bar in some circumstances, especially when Data Boxes were turned on. This occasionally resulted in one part of the tree appearing at a different level than another part.
• If taxon labels in the tree window were any style other than plain (that is, bold, italics, etc.), then the labels for the bars on branches in Trace All Changes mode were also mistakenly of this style.
• If you used the drag-and-drop rerooting tool, and touched on a branch, but did not move the tool off the branch before you let go of the mouse button, MacClade would crash. Bug reported by three users.
• On occasion MacClade would query you to save a tree file when closing a data file, even though no tree file was open.
• On a Macintosh with a 16 or more colors or grays shown, MacClade occasionally incorrectly printed Data Boxes.
• If Trace All Changes is shown, with Minimum and Maximum changes, and you went to Print Tree, and choose Multiple Trees from the Options pop-up menu, then asked to print multiple trees, MacClade would incorrectly show changes on the printed trees, and would give some error messages. Bug reported by two users.
• If you asked to not show the trace legend in Print Tree, this request was forgotten the next time you invoked Print Tree. Bug reported by one user.
• If you asked to print Data Boxes from within Print Tree, printing might not take place until you left the Print Tree dialog box.
• On occasion the preview image in the Print Tree dialog box was not updated correctly. Bug reported by one user.
• If a data file in MacClade 2.1 format contained an invalid tree description, with one taxon occurring more than once, MacClade 3 would usually crash upon reading the file or displaying the tree. Bug reported by one user.
• In Trace All Changes, if the number of changes varied by less than 10 across branches, but the maximum number of changes on any branch is greater than 10, and if the trace was indicated with patterns or colors, then the legend would incorrectly label the patterns or colors.
• In the Character Steps/etc. chart, if one chooses the chart with number of steps on the vertical axis and character number on the horizontal, and changes the interval width to something other than 1 (see pp. 276&endash;277 of the MacClade book), then MacClade may crash if the number of characters is not divisible by the interval width. Bug reported by one user.
• If you modify the Evolve Segments of various branches, then save the file, there are occasions in which the number of segments of the right-most branch on the tree will be incorrectly set on re-opening the file. Bug present only in version 3.01.
• Occasionally a scroll bar will be incorrectly drawn in the textual summary of a State Changes & Stasis chart. Bug present only in version 3.01.
• If a chart is on the screen, and Evolve Characters is invoked, and there are fewer states per character in the simulated characters than in some of the original characters, then the chart may be drawn incorrectly, omitting some of the higher-valued states.
• When an observed taxon is fixed as ancestor then holding the cursor over the corresponding internal branch should cause the message box to indicate the taxon's name, but instead it indicates garbage if the internal branch is a left-leaning branch.
• Whenever the State Changes & Stasis chart was used and observed taxa were fixed as ancestors, the chart gave false results. The results included ficticious changes that varied haphazardly according to situation.
• In printing Char. & State Names from the Print Other submenu, the last character listed on each page may have some states omitted from the list. Bug reported by one user.
• In printing Branch Lists from the Print Other submenu, all pages after the first may have the first line of the page printed at the bottom of the page, thus giving the incorrect impression that a character changes along the branch listed at the bottom of the page, rather than that listed at the top of the page. Bug reported by one user.
• In Trace All Changes, if branch lengths are weighted, and if decimal-valued weights or user-defined types are in use, then in Print Tree, if you choose to show the scale of the branch lengths, then the scale will be labeled incorrectly.

Bugs fixed in 3.03

• On rare occasion, MacClade will refuse to open files from the Open File menu item. This seems to be restricted to hard disks that have been extensively used since last formatted, and that use System 7. Bug reported by six users.
• For Extended Standard or protein data files, if Type Edit opens to display a step matrix type with 8 or less states, and you then display a type with 10 or more states, the scroll bars will overlap part of the step matrix. Bug reported by one user.

Bugs fixed in 3.04

• If Minimum-Average-Maximum Trace All Changes is shown on a tree, and you go to Print Tree and have the tree drawn with branch lengths proportional to number of changes, and ask to print Data Boxes, the taxon names will be printed in the incorrect place. Bug reported by one user.
• If a taxon has footnotes attached to more than one of the characters, then MacClade may fail to print some of these footnotes if they appear near the bottom of a page. Bug reported by one user.
• In the Print Tree dialog box, if you click on the Print Tree button, but then click on the Cancel button to avoid printing, MacClade may perform some unnecessary calculations.
• There were times in which symbols for gaps, missing data, etc., could conflict with those for character states, but MacClade would not provide proper warning. For example, with a DNA data file, you could designate a gap as 'Y', which would conflict with the symbols list used by MacClade for the Genetic Code dialog box, causing it to be displayed incorrectly. Bug reported by one user.
• The CHARSET command in the ASSUMPTIONS block of a NEXUS file was occasionally misread, resulting in either a crash or very sluggish performance. Bug reported by one user.
• If you Save Preferences in the Symbols dialog box for standard or extended standard data files, then set the default data format preference to DNA, RNA, or protein, then any new file you create will be designated as DNA, RNA, or protein, but will have incorrect state symbols.
• In some circumstances, touching just below the column width control in the data editor would cause MacClade to crash.
• Some options in the Trace Labeling dialog box were not saved as preferences properly for Trace All Changes.
• There were some problems in screen display in the Include Exclude Taxa dialog box.
• In exporting polytomous trees to Hennig86, MacClade would incorrectly give a warning that the polytomies in the trees were arbitrarily resolved.
• With a continuous character traced, the trace legend would use as its range for shading the entire range of states on the whole tree. However, the trace would use as its range that on the currently shown clade. Since the ranges determine the assignment of shades/patterns to states, the trace could therefore disagree with the legend when a clade was expanded on the screen. This was fixed by using range on whole tree for both Trace Continuous and Trace All Changes (since adjustment to expanded clade was deemed undesirable).
• If one imported a foreign file format in which some characters were specified as not being of default type for that file format (e.g. a Hennig86 file with some characters being unordered), then when that file was saved within MacClade, the current typeset was occasionally saved incorrectly (e.g. a Hennig86 file may have been saved with all character ordered).
• Switching the number of significant digits for indices such as the retention index was not immediately reflected in the screen display.

Bugs fixed in 3.05

• PHYLIP 3.4/3.5 data files that contained trees were written incorrectly. After the data matrix, but before the trees, a line is supposed to be included containing the number of trees. This line was not written into the files.
• If you tried to import some files as MacClade 1.0 or 2.1 files, when in fact they were files of another format (e.g. PHYLIP), then MacClade would occasionally crash. Bug reported by one user.
• Some files (e.g., in simple table format) in which there were several carriage returns or linefeeds after the last character state symbol could not be imported into MacClade.
• When a taxon was deleted from the matrix, the trees stored in memory were sometimes adjusted incorrectly for the deletion of the taxon if the taxon had been placed as arising out of a polytomous node (and certain other conditions held) [namely, in the polytomy's "ghost" resolution, the taxon is at the base of the polytomous clade.]
• If the number of text characters contained in all of the footnotes stored in a data file exceeded 32,000, the footnotes can become corrupted. Bug reported by three users.
• If Track Cell was turned off in the Edit menu in the data editor, then text characters typed into the cell of a taxon or character name were not displayed immediately. Bug reported by one user.
• The character sets written in TYPESETs, WTSETs, and similar objects in the NEXUS file were occasionally written in a redundant fashion in protein-coding nucleotide sequences.
• Names of user-defined character types that contained a period ('.') were occasionally altered by MacClade.
• Weight sets, type sets, inclusion sets, character sets, and codon positions are saved incorrectly for DNA and RNA sequence data if these sets have odd patterns (for example, if sites 1, 3, 4, 7, 9, 10, 12, 13 were excluded, this would be saved incorrectly, but if every third site were excluded,this would be saved properly). This may lead to incorrect results if the altered assumptions are not noticed. Bug present only in version 3.04. Bug reported by one user.
• If a block of cells is moved toward the start of the matrix in the data editor (using the Block Moves feature), and is moved off the start of the matrix (by holding down the Option key), and if the characters that are moved off of the matrix contain no data (e.g. are all missing), then the remaining cells moved in the block will be reassigned the state present in the highest-num ber charactcer. (We know this sounds complicated, but you would know it if it happened to you!) Bug reported by one user.
• If the tree on the screen has been given a name, and then different branch segments are assigned to the various branches, and the file saved, the branch segment information will be lost when the file is reopened.
• On Mac 660AV and 840AV's, MacClade will on occasion present the user with a -35 error, and will refuse to open many files subsequently. Reported by four users. This is the "Stern AV Bug", which took about 7 months to track down; thanks to David Stern for his heroic efforts in helping us figure this one out.
• If you have user-defined types, and then change the format of the matrix (e.g. from standard data to DNA data) in such a way that the maximum state number in the new matrix is less than that present in the user-defined type, then MacClade may crash if you subsequently choose to edit types.

Bugs fixed in 3.06

• If you read in a NEXUS file that has a data matrix with states represented by numbers other than 0 and 1, but that does not contain a FORMAT statement in a DATA or CHARACTERS block, then MacClade will incorrectly display the states in the data editor using some odd symbols, such as boxes. Bug in versions 3.0-3.05.
• The Search and Replace feature was inadvertently disabled in MacClade 3.05.
• With DNA or RNA sequence data, the program would sometimes save weightsets, typesets, etc., with bogus character numbers (specifically, it would save the weight for a non-existent character). This shouldn't cause any significant problems, other than making the file unreadable by other programs like PAUP without modification. Bug in versions 3.04-3.05.
• There were several instances where MacClade would give incorrect warnings about conflicts between state symbols and symbols used for missing data, gaps, and the matching character symbol. Bug in versions 3.0-3.05.
• In some circumstances modal states were written incorrectly in molecular sequence files. Bug in versions 3.0-3.05.

Bugs fixed in 3.07

• Some problems in importing and exporting PHYLIP files.

Bugs fixed in 3.08

• Some weight sets, type sets, and inclusion sets were incorrectly written by previous versions.
• Crash if one deleted thousands of characters at once.
• PHYLIP files with taxon names right-justified were read incorrectly.
• The keyword "(STEPMATRIX)" was not recognized in a USERTYPE definition in a NEXUS file
• If you used Compress Characters, and a single cell was selected, then the information would be incorrectly transported to another cell.
• There were problems with some punctuation marks if used in taxa, character, or state names.
• Couldn't read state names if the matrix was transposed and MacClade was set to write files in the updated NEXUS format.

Unfixed Bugs

• Occasionally MacClade will give a warning, when going back to the tree window, that MacClade had difficulty in restoring the Evolve Segments. This can occur whether or not you have set the evolve segments. If you see this behavior, please report it to us.
• MacClade 3 cannot read files that are on an MSDOS-formatted floppy disk.

"Suboptimalities" we haven't done anything about yet

• When an already-existing file is saved by MacClade, its "Created" time (as seen in the Get Info window in the Finder) is reset to be the same as its "Modified" time.
• When MacClade is open but no files are open, then double-clicking on a MacClade file in the Finder of System 7 should cause MacClade to read in that file. This does not happen. This is caused by a change in the file management software in Apple's systems.

Other Anomalies

These are "anomalies" that are apparently not MacClade's fault.

• Circle trees are not printed properly on some printers. Apparently this is a bug in some printer drivers. We have noticed this with Laserwriter driver 8.1.1; circle trees print fine with Laserwriter driver 8.2.