STEM 7.4a is fully compatible with models created with the original 7.4 release,
and includes the following updates.
ID
|
Description
|
#3678
|
Circular self-reference not detected when iterating
Issue: If iteration is enabled for formulae in the Editor, then it would
not warn if you accidentally created a pair or series of mutual self references
such as User 1 = User 2 and User 2 = User 1 (where the result is totally arbitrary,
depending only on the previous value of User 2 when the second formula is added).
Action: The Editor now warns if such a direct circularity arises.
|
#3694
|
Indentation of inner fields in tooltips for icons in the Editor
Issue: When you hover over an icon, you get a tooltip summarising all the
values that have been defined within that element, including all the ‘inner fields’
such as time-series parameters and transformation inputs.
Action: The indentation for these inner fields is now restored to a fixed
offset before the value column, irrespective of the indentation of the current field.
|
#3695
|
Stray .icp file created in STEM installation folder when running a replicating model
from Excel
Issue: Due to a quirk of the internal logic, a stray .icp file would be created
in the STEM installation folder when a model which uses template replication was
run from Excel.
Action: This is now fixed.
|
#3696
|
‘Name not found’ message when click help button on add-in toolbar
Issue: Clicking on the help button on the STEM add-in toolbar in Excel would
yield a spurious ‘Name StemPath not Found’ message if there was no workbook open
with a STEM Toolbar definitions sheet.
Action: This is no longer relevant as the button always takes you to the
online version of the help in the latest implementation.
|
#3698
|
Don’t save over a read-only model!
Issue: Previously STEM would rename a read-only .dtl file with
the .~tl
backup extension and then save a new version of the model with the .dtl extension
(and similar behaviour for the .icp file).
Action: STEM now verifies that these files are not read-only before commencing
the backup and save process.
|
#3717
|
Same web-interface ID used for separate model after Save As
Issue: If you had been working with an eSTEM exported interface for Model
1 and then at some point saved a fresh copy of the model as Model 2,
then Model 2 would always inherit the same ID, leading to confusion between the two models
on the server.
Action: A new Generate New ID checkbox in the Create Web Interface dialog
allows you to ask STEM to generate a new ID when required. (It is always set to
false by default.)
|
#3718
|
Unexpected Y0 column in web interface
Issue: A table in an exported eSTEM interface could show a Y0 column when
the original table in the Results program did not.
Action: The web version now matches.
|
#3719
|
Misleading behaviour if duplicate interpolated-series periods or inputs linked from
Excel or ODBC
Issue: The Editor prevents you from adding two separate values for the same
period end (or beginning) in an interpolated series, or for the same input value
in the Mapping field for an Input-Output Mapping transformation, but it is possible
to repeat periods or input values when defining such inputs in Excel. In this eventuality,
STEM would remove the duplicate when displaying the series in the Editor, but would
still receive the raw values if the model were run from Excel.
Action: STEM now displays the linked values as-is in the Editor but checks
for repeat periods or input values before running a model. A duplicate output value
will result in a warning, whereas two inconsistent values will be flagged as an
error.
|
#3721
|
Auto-height calculation for compact web interface fails if inputs section taller
than results
Issue: The auto-height calculation would not make an exported eSTEM presentation
tall enough to accommodate the full inputs section and you would end up with a scrollbar.
Action: The calculation now checks both inputs and results columns.
|
#3724
|
Web interface missing results on first load after updating model configuration
Issue: The set of results and scenarios available from a model running on
an eSTEM server is controlled by the exported file, WebServiceModelConfig.txt; and
a default results set is cached on the server to improve initial load times. The
web service regenerates the cache automatically if the master model file is updated,
but was not checking to see if the config had been updated. Thus there could be
results missing in a newly-exported web presentation when first loaded.
Action: The web service now checks and regenerates the cache if the config
has been updated since the cache was created.
|
#3725
|
Link to exported model data with relative path
Issue: An old issue had re-appeared where a new link to an Excel file created
after exporting model data was established with an absolute file path (including
drive and folder) even though the export workbook is created in the same folder
as the model by design.
Action: The intended behaviour is restored such that an external link is
always created with a relative path if possible (if located in the same folder or
a sub-folder relative to the model).
|
#3731
|
Wrong output when arranging graphs for printing
Issue: Some graph attributes were not printed correctly when using the
Arrange Graphs ... option in the Print dialog in the Results program.
Action: The reported issues have been fixed.
|
#3732
|
Missing output when printing model data
Issue: Due to an unforeseen side-effect of other formatting changes for tooltips,
the print output was missing field labels.
Action: The correct output has been restored.
|
#3736
|
Alternative scenario name not used in web interface
Issue: Some charts in an exported eSTEM interface would always show the logical
name for a scenario (e.g., Variant 1/Variant 2), rather than an alternative name
added with the new Rename Scenario command in the Scenarios and Sensitivities dialog
in the Editor (introduced in STEM 7.4).
Action: The alternative name is now used throughout.
|
#3738
|
Working model not run even though Include Working Model is checked
Issue: Immediately after doing a Save As, STEM would ignore the
Include Working Model flag until the next time the control was unchecked and re-checked.
Action: This was an unintended side-effect of resetting the action of <F5>
when saving a model with a new name. The logic has been clarified to just revert
the association of <F5> to Run Working Model without altering the status of
the Include Working Model flag for what happens if you subsequently Run Selected
or All Scenarios.
|
#3739
|
Finding individual inputs and results in the new online help
Issue: The new online help resource has a comprehensive search function,
but it was not very good at finding the principal descriptions and definitions of
inputs and results as these are presented in longer, alphabetical listings which
combine the descriptions for all items beginning with the same letter, or parallel
pages listing all inputs in a given data dialog.
Action: The individual descriptions and definitions are now indexed in their
own right and associated with suitable bookmarks in the various alphabetical-listing
and data-dialog pages. The net result is that, if you search for an input or result
name, these bookmarks should come out on top of the search results. Tip: in most
cases this will be sufficient; but, if a term is more widely used, you can focus
the search by adding the words ‘input’ or ‘result’ to the search query.
|
#3740
|
Strange automatic-magnitude setting for table with no values
Issue: A table in the Results program would show a funny magnitude scaling
on the value axis if none of the displayed results were available (i.e., up to date).
Action: The magnitude scaling is set to None if no values are available for
the usual calculation.
|
#3741
|
Help on Current Window not the same as <F1> in the Results program
Issue: If you press <F1> in the Results program, it shows help on the
result(s) currently graphed, unless a dialog is open (in which case it shows help
for the dialog). If no graphs have been drawn, then generic help on opening results
is displayed. The Help menu has a Current Window item which is supposed to be equivalent
to pressing <F1>, but it would always go to the generic topic.
Action: The inconsistency was a by-product of the transition from the previous
WinHelp implementation and the logic has now been fixed.
|
#3742
|
Clipped brackets in Scenarios and Sensitivities dialog
Issue: At some display resolutions, insufficient height was allocated per
row for the text to be displayed completely in the list of scenarios. Most notably
the square brackets used to indicate hidden scenarios could be slightly clipped.
Action: The display logic has been simplified to avoid this problem.
|
#3743
|
Grid for interpolated-series input too wide in web interface
Issue: If a web presentation generated by eSTEM is embedded within a master
page layout which imposes a fixed size on the model ‘container’, then the various
controls and charts are intended to scale to fit. Below a certain size, a very slight
over-estimate of the width for individual cells in an interpolated-series grid would
make the grid as a whole too big and result in the container gaining an unsightly
scrollbar.
Action: The calculation has been refined to avoid this outcome.
|
#3746
|
Error message when Move Up from tabular dialog
Issue: A baffling but actually benign internal error message was reported
if you selected Up from the Move menu in a tabular dialog in the Editor.
Action: The internal logic has been fixed. Note: it is not possible to Move
Up from a selection dialog such as you would access from the icon menu when several
elements are selected.
|
#3749
|
Consistent crash with Auto Graph function in the Editor, but only on Windows 8
Issue: A late-stage and seemingly harmless change to one of the library modules
in our graphing system (focused on behaviour in the Results program) had the unanticipated
side-effect of breaking the Auto Graph function, such that it would consistently
crash the Editor. Somehow this managed to slip through the net and was not detected
in routine, pre-release testing. The issue came to light when we were updating our
training materials in January, since when a simple patch has been available.
Action: The issue is now resolved.
|
#3750
|
Gap in documentation for the Maximum Credit input for a Debt Facility element
Issue: The Debt Facility element allows you to model specific financial instruments
such as may be provided by a bank or a vendor to provide funding for the business
being modelled. The Maximum Credit input is self-explanatory and is designed to
provide no constraint if set to zero. However, this detail was not covered in the
documentation.
Action: The online documentation was updated in situ some weeks ago and the
latest content will be installed offline with STEM 7.4a.
|
#3753
|
External link not updated during the session when first added
Issue: It emerged during training that, immediately after establishing an
input link from Excel, STEM would subsequently appear not to notice at run-time
if the Excel file had been saved more recently and would still run with the previous
values. Internally, the external link has a ‘referenced’ flag which, as a consequence
of other recent changes, was no longer initialised correctly in this situation.
STEM would notice that the file had changed, but not realise that anything depended
on it.
Action: The external link is now correctly initialised and STEM will prompt
to update subsequent changes as usual.
|
#3754
|
Difference chart comes back with wrong base scenario after reload results
Issue: A difference chart comparing various scenario results to one other
specific scenario could come back with the base scenario changed to the working
model in a particular set of circumstances, depending on exactly which scenarios
were up to date when the workspace was saved.
Action: The logic is now fixed such that the base scenario identity is always
saved correctly.
|
#3755
|
View content lost in the Results program
Issue: STEM 7.4 makes results views easier to work with by making the previous
fixing process optional: the content and arrangement of a regular view is now automatically
preserved as you step between different views. Unfortunately this logic would appear
to fail if the results were updated between the graphs being drawn and subsequently
changing to another view.
Action: The logic is now fixed such that updating the results will not interfere
with the management of changes to individual views.
|
#3756
|
Charts sometimes get bigger in web interface after model is run for the first time
Issue: Our exported eSTEM interface logic puts the STEM content within a
div with class stem_autosize_height which triggers a client-side script to size
the div to the required content and thus remove any unsightly scrollbars. However,
this happens after the charts are first drawn in the original available space (to
calculate the required height, depending on font sizes and so on), and they would
then be redrawn slightly wider the next time the model was run if the div had been
resized (expanding into the space vacated by the scrollbar).
Action: The charts are now redrawn immediately if the size of the div is
changed by the auto-size script.
|
#3757
|
N/A in time-series table when enter new value for existing period
Issue: The Mappings dialog and any other time-series table displays values
for every period of the model run. If the underlying data (accessed from the first
button in each row) is an interpolated series, then values for periods of the interpolated
series which coincide with a period of the model run (according to the appropriate
beginning or end alignment) are displayed in black in the time-series table. Such
values can be edited directly to modify the underlying data.
However, if the underlying period is of a different length to the period of the
model run (e.g., quarter compared to year), then a new and effectively duplicate
period would be added (e.g., year, in addition to the existing quarter) which would
in turn lead to a N/A in the time-series table.
Action: The value for the existing, underlying period will be modified now
to avoid this undesirable duplication.
|
#3758
|
Copy and paste from Excel fails with thousands-separated formats
Issue: It is usually possible to copy a 2xn block of cells from Excel, combining
periods and values, into an interpolated-series dialog in the Editor. However the
data would be truncated if originally displayed in Excel with a thousands separator.
Action: This now works correctly in the majority of cases where Excel is
respecting the operating-system regional settings for thousands separators.
Notes: We have seen isolated cases where the Excel set-up has been incorrect.
Because this is beyond our control, we are working on using what would seem intuitively
to be the more robust ‘binary interchange’ format from the clipboard.
However, this is not as straightforward as you might think and we will need to do
more research and testing before we will be ready to use this in preference to the
current text format.
|
#3761
|
Certain value-axis settings not reflected in web interface
Issue: The minimum, maximum, major unit and minor unit defined for the value
axis of a chart in the Results program had no impact on an exported eSTEM interface.
Action: These settings now flow through as expected.
|
#3762
|
Error message not displayed properly for template instance
Issue: When a model runs with scenarios or with template replication, errors
relating to specific variant data may not be readily apparent in the working model.
Therefore the expanded model for each scenario is checked independently. Any error
messages will identify the relevant scenario and/or template instance and highlight
the relevant location in the model and associated variant data. This complex arrangement
could be compromised if a replicated element already had a ‘/’ in its
name; an error would still be reported for the correct element but the actual gist
of the message would be lost.
Action: The logic is no longer sensitive to special characters in element
names and the error message is now displayed correctly in the original situation
reported.
|
#3766
|
Stacked format lost when show as separate table or graph
Issue: The stacked aspect of a graph or table would be lost when shown as
a separate table or graph.
Action: The stacked aspect is now preserved.
|
#3767
|
Incorrectly maximised table in the Results program
Issue: The Results window would not be redrawn properly if you switched to
a view or loaded a workspace including a maximised table.
Action: The display logic is fixed.
|
#3768
|
Recent file list missing entries after manual edit
Issue: If you work with STEM as a consultant, sometimes it can be pertinent
to purge certain entries from the recent file list in the Editor (or Results program)
to avoid making visible the names of models relating to other clients. As the code
was originally, if you manually removed one entry from the middle of the list (stored
in STEM.INI) then all subsequent entries would disappear from the recent file list
too. More confusingly, these later entries might come back later once the missing
entry was filled in by current file activity.
Action: Now all remaining items in STEM.INI will appear on the menu. The
next time the list is updated all current items will be recorded ‘without
gaps’.
|
#3769
|
Tooltip for icon obscures scrollbar in the Editor
Issue: If you tried to use the scrollbar for a view in the Editor, a tooltip
might be displayed for an icon at the edge of the view behind the current mouse
position over the scrollbar.
Action: A tooltip will no longer appear if the mouse is over a scrollbar.
Moreover, any existing tooltip will be removed if you move the mouse over a scrollbar.
|
#3772
|
Ellipsis on Rename item in data-dialog context menu
Issue: In general terms, menu items are supposed to provide some clue as
to the immediacy of their action by including an ellipsis (…)
if the menu item leads to a dialog or message box where further input will be required
before the action itself. For example, on an icon menu, all the items which lead
to data dialogs are immediate, whereas Delete… and Rename… prompt for
confirmation. A sharp-eyed user spotted that the corresponding ellipsis was missing
on the Rename item on the new context-menu in data dialogs.
Action: Uniquely on this context-menu, the Rename item now has an ellipsis
too.
|
#3773
|
Ambiguous reference message made more specific!
Issue: When you create a formula in the Editor, it is possible for a reference
to another field to be ambiguous if the model contains an element or user data,
the name of which (e.g., Tariffs) matches the name of an internal field in the element
where the formula is being defined.
Action: STEM issues a warning in this situation and the message now identifies
the conflicting interpretation as well as the one which is currently assumed.
|
#3774
|
What used to be known as the Select button is not the same as Find
Issue: When the new Find function was added in STEM 7.4, the tooltip for
the first button on the toolbar was changed to say, ‘Finds and displays a model
element’. The corresponding item on the Element menu was changed to say Search.
However the corresponding dialog lists elements by type and allows you to add icons
to the current view rather than find existing icons.
Action: To avoid confusion with the Find command, the tooltip now says, ‘Lists
and displays model elements’, and the menu item is now List. The dialog caption
itself is now List Elements.
|
#3775
|
Simpler messaging for changes to fixed views
Issue: As well as making the fixed behaviour of views in the Results program
optional, STEM 7.4 also provides the option to prompt when leaving a fixed view
which has been modified (e.g., by re-arranging or drawing new graphs) so that you
can choose whether to keep the changes. If you switch to another view, the message
is simply, “Retain changes to fixed view ‘View 1’?” In contrast, a much
more complicated message was issued in the case where you would select the current
view from the View menu.
Action: Now the message has been re-worded as, “Revert to previously
fixed appearance of view ‘View 1’?”
|
#3778
|
Making it easier to find the other end of a link in the Editor
Issue: In a large view, it could be fiddly if you had to scroll to find out
what is at the other end of a link to or from an icon.
Action: A new Show Other End command on the context menu for the link selects
and scrolls into view the icon at the other end (i.e., furthest from where the context
menu originates).
|
#3776
|
Edit or just review details corresponding to a link between two icons
Issue: The first item of the context menu for any of the automatic links
between icons in the Editor was labelled Edit, meaning show the relevant dialog
where the link is defined.
Action: The menu item is now labelled Details to be consistent with the
simpler icon menus, reflecting the fact that you may wish only to review rather
than necessarily change the settings.
|
#3779
|
Copying element names and text from a view
Issue: If you copy a selection of icons in the Editor then the corresponding
element names are put on the Windows clipboard as text.
Action: The text from text boxes is now presented with the same tab-separated
format for consistency when pasting into Excel.
|
#3780
|
Understanding the Show in a New View button in the Find dialog
Issue: The Show in a New View button in the Find dialog is designed to only
apply to elements in the results list as a text box could only be shown in a new
view by copying it.
Action: A message is now displayed in the status bar if you click this button
when no elements are selected.
|
#3781
|
<Esc> doesn’t work in zoom mode
Issue: The zoom window (top-left corner of a view in the Editor) can be used
to navigate around the associated view, simply by clicking and dragging the inner
rectangle of the zoom window which represents the visible portion of the view. If
you change your mind while doing this, you should be able to press <Esc> as
you can during other mouse-drag operations such as moving icons, copying icons or
rubber-band selecting.
Action: Now you can also press <Esc> to cancel zoom mode and return
the view to its original position without modifying the layout.
|
#3785
|
Tooltip on inactive data-dialog caption obscures current window
Issue: If you accessed the Constant dialog for a field in a data dialog in
the Editor, but then left your mouse hovering over the caption of the original dialog,
after a few seconds the presence of the mouse would initiate a tooltip for the parent
dialog which would in turn come on top and obscure the Constant dialog you were
trying to edit!
Action: The presence of the tooltip no longer brings the associated window
to the top.
|
#3788
|
Fixed margin below graphs in a web interface unsuited to some layouts
Issue: The original eSTEM layout code would always add a fixed margin of
40 pixels below each graph in a generated web presentation. This was not very compatible
with the new compact layout option and was more or less impossible to customise.
Action: Now the margin is defined as part of the stylesheet in such a way
that it can be easily overridden if necessary in userstyles.css.
|
#3795
|
Relative Path checkbox in Define External Link dialog should be checked by default
Issue: If you used the Add button in the External Links dialog in the Editor
to add a reference to an external file, the file would not be stored with a relative
path by default, even if the file was in the same folder as the model.
Action: The Relative Path checkbox is now checked by default and will stay
checked after you use the Browse button to select a file so long as the file is
in the same folder as the model (or a sub-folder).
|
#3798
|
Offline help content not launched properly on some computers
Issue: STEM 7.4 would use Windows file associations to decide how to open
the HTML content, but this could lead to it being opened in a text editor in some
cases.
Action: Help is now opened with reference to http protocol associations which
should more specifically and reliably locate your favourite web browser.
|
#3803
|
STEM web interface chokes on very large interpolated series
Issue: The size of an update request received from the web (including all
required input values) is controlled for security reasons. However, the original
limit of 1000 characters proved too small for generated interfaces with more than
a certain number of periods across multiple interpolated-series inputs.
Action: The limit has been increased to 10,000 characters and is now further
configurable on the server.
|