1:56 a.m.
In case anyone's wondering, "Where did that menu item go?", here's
an account of the changes I've made in the main window menus
lately (some in response the comments of Talha and others, some
prior to that). (This is all without prejudice to any larger
scheme for reorganizing things, but might be helpful if we do
decide to go that way.)
Data menu:
- Consolidated "Read info" and "Edit info" into "Dataset
info"
(you just get to read if it's a packaged data file; you can edit
if it's your own data file).
- Consolidated "Add markers" and "Remove markers" into
"Observation markers". The choices you get depend on whether the
dataset has observation markers or not; it it does, you get a new
option to export the markers to file (enabling you to "clone" the
markers from one dataset to another).
- Moved the items "Set missing value code" and "Count missing
values" to here from the Sample menu; moved the item "Find
variable" here from the Variable menu.
- Removed the old "Print report" item as redundant.
Sample menu:
- Consolidated the items "Restrict, based on dummy" and "Restrict,
based on criterion". Using a dummy is now an option in the "based
on criterion" dialog, if there are dummies available.
- Moved out items that are not specifically related to the sample
setting (see the Data menu).
- Moved the item "Show status" to the end, as the odd man out
among the entries.
Variable menu:
- Moved up "Set missing value code" so it's not beneath the time
series-only options.
- Moved out all items that are not specifically related to the
currently selected variable (see Data menu).
- Consolidated the "Frequency distribution" and "Frequency plot"
items. (The Frequency distribution dialog now has a checkbox for
showing the plot.)
Tools menu:
I experimented today with a couple of the suggestions made by
Talha, namely moving the Working Directory dialog and the font
selectors from positions in the menu structure into the
Preferences dialog box. I have now reverted both of these
changes. I reverted the working directory move on account of
Sven's objection that this is not a "relatively settled" choice
for some users. I reverted the font selection items because the
move gave rise to a confusing situation with two levels of buttons
("OK"/etc.), the relationship between which was unclear.
Allin Cottrell
--
Allin Cottrell
Department of Economics
Wake Forest University
8:18 a.m.
Sorry for acting a bit impetously earlier. I do like the new menu
changes. They definitely give a cleaner and more consistent look. Here
is a bunch of comments and suggestions. I hope you take these into
consideration also:
1)- How about renaming all the "... data" items (open data, append
data, save data, etc.) items in the file menu simply open, append,
save, save as, etc. (and "Clear data set" will be renamed close)? This
would be an improvement because we already have a "data" menu and
secondly this way is consistent with the naming scheme of other
programs (openoffice, gnumeric, etc.).
2)- How about consolidating the "Add observations" and "Remove extra
observations" items into one "Dataset size" item next to the "Dataset
structure" item?
3)- How about consolidating the "Display values" and the "Edit
values"
items in the Data menu into one "Edit values" item which will give the
spreadsheet for the whole data set as in the "Data set" item in the
icon view?
4)- Modern programs do not have a "Working directory" setting because
the various open and save dialog boxes remember the last directory. I
for one find it rather annoying that gretl does not do this. If
implemented, I think Sven would appreciate this feature as well.
5)- What is the relationship between the two "Set missing value code"
items in the Data and Variable menus? I find this rather awkward and
confusing. What will happen if one sets a NA code for a single
variable and then chooses something else from the data menu? Is it a
good idea to have different missing value codes for different
variables? Plus, it is a fundamental GUI design principle that there
should not be duplicate menu items.
6)- RNG seed is not a tool.
7)- Spearman's rank correlation is not a model. How about putting this
as a checkmark into the "Correlation matrix" dialog? Instead of having
the rho value computed each time for two variables, we would have a
nice matrix of all the values. This is what Eview does.
Best regards
Talha
On Thu, Dec 3, 2009 at 3:56 AM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
In case anyone's wondering, "Where did that menu item go?", here's
an account of the changes I've made in the main window menus
lately (some in response the comments of Talha and others, some
prior to that). (This is all without prejudice to any larger
scheme for reorganizing things, but might be helpful if we do
decide to go that way.)
Data menu:
- Consolidated "Read info" and "Edit info" into "Dataset
info"
(you just get to read if it's a packaged data file; you can edit
if it's your own data file).
- Consolidated "Add markers" and "Remove markers" into
"Observation markers". The choices you get depend on whether the
dataset has observation markers or not; it it does, you get a new
option to export the markers to file (enabling you to "clone" the
markers from one dataset to another).
- Moved the items "Set missing value code" and "Count missing
values" to here from the Sample menu; moved the item "Find
variable" here from the Variable menu.
- Removed the old "Print report" item as redundant.
Sample menu:
- Consolidated the items "Restrict, based on dummy" and "Restrict,
based on criterion". Using a dummy is now an option in the "based
on criterion" dialog, if there are dummies available.
- Moved out items that are not specifically related to the sample
setting (see the Data menu).
- Moved the item "Show status" to the end, as the odd man out
among the entries.
Variable menu:
- Moved up "Set missing value code" so it's not beneath the time
series-only options.
- Moved out all items that are not specifically related to the
currently selected variable (see Data menu).
- Consolidated the "Frequency distribution" and "Frequency plot"
items. (The Frequency distribution dialog now has a checkbox for
showing the plot.)
Tools menu:
I experimented today with a couple of the suggestions made by
Talha, namely moving the Working Directory dialog and the font
selectors from positions in the menu structure into the
Preferences dialog box. I have now reverted both of these
changes. I reverted the working directory move on account of
Sven's objection that this is not a "relatively settled" choice
for some users. I reverted the font selection items because the
move gave rise to a confusing situation with two levels of buttons
("OK"/etc.), the relationship between which was unclear.
Allin Cottrell
--
Allin Cottrell
Department of Economics
Wake Forest University
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
“Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment.” - Benjamin Franklin (1706-1790)
--
9:56 a.m.
Talha Yalta schrieb:
Sorry for acting a bit impetously earlier.
Talha, I do take your proposals seriously and as I said before I am
grateful that you draw up a reform concept. But reacting to it in a
coherent way also takes time. Personally I will have time to think about
it next week (starting Sunday). So instead of offering more isolated
reactions I suggest to give other interested people some time, which you
shouldn't then interpret as dis-interest. Of course I'm not saying that
the discussion has to stop until then, but maybe I would humbly ask that
no premature decisions are made.
thanks,
sven
9:27 p.m.
First of all, Like Sven I won't be able to do much in this area in
the next couple of weeks, but I'll be interested to hear more
reactions to recent changes and proposals. Meanwhile I'll respond
to some of Talha's points.
On Thu, 3 Dec 2009, Talha Yalta wrote:
1)- How about renaming all the "... data" items (open data,
append
data, save data, etc.) items in the file menu simply open, append,
save, save as, etc. (and "Clear data set" will be renamed close)? This
would be an improvement because we already have a "data" menu and
secondly this way is consistent with the naming scheme of other
programs (openoffice, gnumeric, etc.).
Gretl is very different from those other programs. We have to
handle various disparate types of files, including but not limited
to data files. I can't see any advantage in removing the
information that the files referenced in the top half of the File
menu are data files. And "Clear data" is much more descriptive
than "Close".
3)- How about consolidating the "Display values" and the
"Edit values"
items in the Data menu into one "Edit values" item which will give the
spreadsheet for the whole data set as in the "Data set" item in the
icon view?
Displaying and editing data are different operations with
different associated options. Non-starter. Plus with a big
dataset you almost certainly don't want to open the whole thing at
once in the data editor -- that's a very resource-intensive
operation.
4)- Modern programs do not have a "Working directory"
setting because
the various open and save dialog boxes remember the last directory. I
for one find it rather annoying that gretl does not do this. If
implemented, I think Sven would appreciate this feature as well.
Modern, schmodern. Gretl requires and supports more sophisticated
directory/path handling than "pure GUI" programs. Files can be
opened and saved via script and console as well as via GUI
dialogs. Experts will often start gretl from the command prompt,
not via an icon, and we want proper reversion to the current
working directory (unless we don't, in which case we can turn it
off).
However, I have made one modification here. In the Working
directory dialog there's now one more binary choice: to have the
GUI file selector remember and revisit the directory last used for
an Open/Save operation, or always to start in the selected working
directory. Note that this affects the GUI selector only; and it
doesn't persist across sessions. I mean, the preference persists,
but the name of the "last visited folder" does not.
5)- What is the relationship between the two "Set missing value
code" items in the Data and Variable menus?
Try clicking the little button with "Help" written on it ;-)
This is about fixing data that has been read in from some third
party source. One chooses a value -- -1, 999, whatever -- and
says that either for a particular variable, or for all variables,
that value should be converted to gretl's internal NA.
What will happen if one sets a NA code for a single variable and
then chooses something else from the data menu?
Choose -1 for a particular variable; all -1s become NAs.
Choose 999 for the dataset: all 999s become NAs. No problem.
6)- RNG seed is not a tool.
If it's not a tool (I still say it is), then a tool is the thing
it most resembles. It is certainly not a "preference". I don't
"prefer" 12067543, I choose to set it on a particular occasion.
Allin.
7:45 a.m.
Thanks very much for your useful comments. I have to agree with all of
your points and I am absolutely grateful for making the GUI selector
remember the last directory. The old behavior was so annoying for me
that I think my life expectancy has increased a bit with this :-)
I have just a few quick comments:
1)- You haven't commented on my points 2 and 7. Does this mean you
don't disagree with them? \(^_^)/
2)- Remebering the last directory is so great and natural that I don't
think anybody would disagree if it was the standard behavior instead
of an option. If the user prefers working in the working directory the
program remembers that too, no? Plus, the working directory can be
changed at any time.
3)- I think the file menu could be improved (become simpler and more
standard) if you could remove the import data sub-menu and add this
functionality to the open user file dialog box as a drop-down "file
format" option. Similarly, the "Export data" sub-menu would be an item
where the file format choice is done within the GUI selector as most
other programs do.
4)- I still think that there must be a better way to do the missing
value function. How about gretl doing the fixing data aspect of it at
the time of opening the data?
5)- I would like to point out that I do support (with Berend) having a
"Window" menu even if it will be the 10th menu. The only down side I
can think of is that this may require the main window getting a bit
wider by default in some translations.
Cheers
Talha
On Thu, Dec 3, 2009 at 11:27 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
First of all, Like Sven I won't be able to do much in this area in
the next couple of weeks, but I'll be interested to hear more
reactions to recent changes and proposals. Meanwhile I'll respond
to some of Talha's points.
On Thu, 3 Dec 2009, Talha Yalta wrote:
> 1)- How about renaming all the "... data" items (open data, append
> data, save data, etc.) items in the file menu simply open, append,
> save, save as, etc. (and "Clear data set" will be renamed close)? This
> would be an improvement because we already have a "data" menu and
> secondly this way is consistent with the naming scheme of other
> programs (openoffice, gnumeric, etc.).
Gretl is very different from those other programs. We have to
handle various disparate types of files, including but not limited
to data files. I can't see any advantage in removing the
information that the files referenced in the top half of the File
menu are data files. And "Clear data" is much more descriptive
than "Close".
> 3)- How about consolidating the "Display values" and the "Edit
values"
> items in the Data menu into one "Edit values" item which will give the
> spreadsheet for the whole data set as in the "Data set" item in the
> icon view?
Displaying and editing data are different operations with
different associated options. Non-starter. Plus with a big
dataset you almost certainly don't want to open the whole thing at
once in the data editor -- that's a very resource-intensive
operation.
> 4)- Modern programs do not have a "Working directory" setting because
> the various open and save dialog boxes remember the last directory. I
> for one find it rather annoying that gretl does not do this. If
> implemented, I think Sven would appreciate this feature as well.
Modern, schmodern. Gretl requires and supports more sophisticated
directory/path handling than "pure GUI" programs. Files can be
opened and saved via script and console as well as via GUI
dialogs. Experts will often start gretl from the command prompt,
not via an icon, and we want proper reversion to the current
working directory (unless we don't, in which case we can turn it
off).
However, I have made one modification here. In the Working
directory dialog there's now one more binary choice: to have the
GUI file selector remember and revisit the directory last used for
an Open/Save operation, or always to start in the selected working
directory. Note that this affects the GUI selector only; and it
doesn't persist across sessions. I mean, the preference persists,
but the name of the "last visited folder" does not.
> 5)- What is the relationship between the two "Set missing value
> code" items in the Data and Variable menus?
Try clicking the little button with "Help" written on it ;-)
This is about fixing data that has been read in from some third
party source. One chooses a value -- -1, 999, whatever -- and
says that either for a particular variable, or for all variables,
that value should be converted to gretl's internal NA.
> What will happen if one sets a NA code for a single variable and
> then chooses something else from the data menu?
Choose -1 for a particular variable; all -1s become NAs.
Choose 999 for the dataset: all 999s become NAs. No problem.
> 6)- RNG seed is not a tool.
If it's not a tool (I still say it is), then a tool is the thing
it most resembles. It is certainly not a "preference". I don't
"prefer" 12067543, I choose to set it on a particular occasion.
Allin.
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
“Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment.” - Benjamin Franklin (1706-1790)
--
2:35 a.m.
On Fri, 4 Dec 2009, Talha Yalta wrote:
Thanks very much for your useful comments. I have to agree with all
of
your points and I am absolutely grateful for making the GUI selector
remember the last directory. The old behavior was so annoying for me
that I think my life expectancy has increased a bit with this :-)
Well, I'm glad to hear that!
I have just a few quick comments:
1)- You haven't commented on my points 2 and 7. Does this mean you
don't disagree with them? \(^_^)/
Just haven't arrived at a definite view yet.
2)- Remebering the last directory is so great and natural that I
don't think anybody would disagree if it was the standard
behavior instead of an option.
It is in fact the default at present.
3)- I think the file menu could be improved (become simpler and
more standard) if you could remove the import data sub-menu and
add this functionality to the open user file dialog box as a
drop-down "file format" option. Similarly, the "Export data"
sub-menu would be an item where the file format choice is done
within the GUI selector as most other programs do.
I agree with that, but don't expect it to happen right away.
4)- I still think that there must be a better way to do the
missing value function. How about gretl doing the fixing data
aspect of it at the time of opening the data?
We do our best to get this right on opening the data, already.
Perhaps fixup of misrecognized NAs is too arcane to have in the
GUI. What do others think? It's easy enough to do the fixup, if
required, in script mode, as in:
scalar badval = -1
xvar = (xvar == badval)? NA : xvar
But the reason this item is present at all is that some of the
sorts of data we can import (sas, stata, spss?) have a slew of
special missing value codes which can't always be handled
perfectly.
5)- I would like to point out that I do support (with Berend) having
a
"Window" menu even if it will be the 10th menu. The only down side I
can think of is that this may require the main window getting a bit
wider by default in some translations.
Try running gretl in German.
Allin.
10:38 a.m.
Allin Cottrell schrieb:
> 5)- I would like to point out that I do support (with Berend) having a
> "Window" menu even if it will be the 10th menu. The only down side I
> can think of is that this may require the main window getting a bit
> wider by default in some translations.
Try running gretl in German.
as the German translator, I wouldn't want to be a showstopper. Maybe the
German menu names could be shortened somewhat (though it's not so
trivial, really, if the names are supposed to be meaningful translations
-- German-speaking list members, I'm glad to hear your suggestions to
replace 'Werkzeuge', 'Hinzufügen', 'Stichprobe'!).
But I also would like to clarify that I'm a bit worried about the sheer
number, not so much the total width of the menu bar.
thanks,
sven
4:36 p.m.
On Sat, 5 Dec 2009, Sven Schreiber wrote:
as the German translator, I wouldn't want to be a showstopper.
[for adding an additional top-level menu item]
Don't worry, you're not. My comment was somewhat tongue-in-cheek.
But I also would like to clarify that I'm a bit worried about
the sheer number, not so much the total width of the menu bar.
Agreed here.
I myself floated the idea of consolidating (somehow) the Data,
View and Sample menus -- reducing 3 to 2 perhaps -- and if that
were possible it could make room for another entry, but on
reflection (and after consolidating several items under those
headings) I'm not enthusiastic about this.
Having moved some items, I think the Data and Sample menus are now
well focused. I would not want to give up the Sample menu, because
setting the sample is a basic and common task; it shouldn't be
buried under Data, IMO (plus it would be nastily backward
incompatible to "hide" it).
The View menu is somewhat miscellaneous, but we're talking the art
of the possible here. Solely on the criterion of "clear focus" of
menu headings we'd have to divide this into three headings:
1) "Windows" (or some such)
2) "Graphs"
3) "Multivariate statistics that are not quite 'models'".
That's obviously a non-starter. I think you just have to get used
to what's under gretl's View menu. With any complex program, you
have to "get used to" the menus to some extent; they can be better
or worse designed but they can never be totally transparent.
Then, OK, finally, the (top-level) "Window(s)" menu. Having
experimented, I'm now against it. I don't think it earns its
keep. Most of the GUI software I use regularly doesn't have it:
Emacs - no (duh!)
Firefox - no
Gimp - no
Inkscape - no
Gnumeric - no (it's under View, as with current gretl)
OpenOffice - yes, but it's aiming to be an MS-lookalike.
Berend mentions OS X. I agree that if we ever get cosy enough
with OS X to populate the top-of-screen application menubar,
then that should have a Windows item since it's very standard.
But that's a different issue, IMO.
Allin.
5:12 p.m.
I agree with you after the recent menu changes. Things look better,
more focused etc. But please do not stop here. While you are at it,
please consider:
1)- Consolidating a add obs and remove extar obs in the data menu and
if possible some items in the variable menu (gini coeff item, any
one?). Those are the two biggest menus now. The variable menu is still
taller than the default height of the main window.
2)- Adding a Spearman's correlations as a matrix,
3)- Import and export data as parts of the GUI selector,
4)- Figuring something about the two "missing value code" items.
5)- Multiple model tables etc.
Recent experience has shown me that GUI is very important and new
users tend form a strong first impression after using gretl for a very
short period of time. If anything looks weird or unprofessional or if
something seems broken, this really negatively affects the perception
of the overall quality. By contrast, GUI items like model tables,
LaTeX export functionality or a possible new gui based matrix
operation functionality provide a significant "wow" factor.
My point is: There are things that need you magic touch so please go
ahead with the GUI improvements at least in this release.
Cheers
Talha
--
“Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment.” - Benjamin Franklin (1706-1790)
--
8:23 p.m.
On 05-12-2009, at 17:36, Allin Cottrell wrote:
Then, OK, finally, the (top-level) "Window(s)" menu. Having
experimented, I'm now against it. I don't think it earns its
keep. Most of the GUI software I use regularly doesn't have it:
Emacs - no (duh!)
Firefox - no
Gimp - no
Gimp 2.6.7 on OSX running with X11 has a Windows top menu.
Gimp 2.6.7 on Ubuntu Karmic also has it.
Berend
8:56 p.m.
On Sat, 5 Dec 2009, Berend Hasselman wrote:
Gimp 2.6.7 on OSX running with X11 has a Windows top menu.
Gimp 2.6.7 on Ubuntu Karmic also has it.
I wouldn't call Gimp the wunderkind of usability. ;-)
Sorry to jump in the GUI debate at this stage: I'm not overly bothered
really, and I doubt I could provide any insightful contribution, since I
use gretlcli most of the time. However, I don't recall workspaces being
mentioned. The way I personally work, if gretl opens a gazillion windows,
no big deal: I just send those I need to a different workspace, and voilà
(rotating cubes and all, for those who like eye candy). I understand
modern windows version have workspaces too. Don't know about the Mac,
though.
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
10:34 p.m.
Sorry, for maybe stupid question.
I've just downloaded the source form cvs and tried to build it for the
first time.
And it seemed that all went right and with no errors until I have
tried to launch the bin/gretl (is it correct? bin/gretl_x11 provides
me with the same result) and the window is without the menu bar. It
looks like this:
http://picasaweb.google.com/moradan228/Something#5411880093204622882
Is it because of discussing now changes in the menu?
My system is Ubuntu 9.10 (with a not very suitable 1.8.1 gretl out of
the box at presence of 1.8.6) and the problem may be in the
dependencies - there were some missing dependencies from recommended
by the manual and some were with much newer version than recommended
(e.g. ligfortran3 instead of libgfortran1). I expected that missing
critical dependency there will be an error during building but there
were no.
5:01 a.m.
On Sun, 6 Dec 2009, Ivan Sopov wrote:
I've just downloaded the source form cvs and tried to build it
for the
first time.
And it seemed that all went right and with no errors until I have
tried to launch the bin/gretl (is it correct? bin/gretl_x11 provides
me with the same result) and the window is without the menu bar. It
looks like this:
http://picasaweb.google.com/moradan228/Something#5411880093204622882
Is it because of discussing now changes in the menu?
No, it's just that you have to install gretl to get the full GUI
in place.
Allin
5256
days inactive
5259
days old
12 comments
6 participants
participants (6)
-
Allin Cottrell -
Berend Hasselman -
Ivan Sopov -
Riccardo (Jack) Lucchetti -
Sven Schreiber -
Talha Yalta