5:30 p.m.
Hi,
it seems to me that gretl on windows is not as "snappy" as it used to
be. I'm referring at least to 1.8.7 and the current snapshot, not sure
when I started seeing the behavior, as I work with gretl more often on
Linux.
I experience short but annoying delays when windows are opened and stuff
like that. I'm guessing the GTK toolkit is responsible -- I seem to
remember that an updated version has been used by gretl recently?
Anybody else have noticed it?
thanks,
sven
5:38 p.m.
Sven Schreiber schrieb:
Anybody else have noticed it?
sorry for the email mini-flood, I was just reminded of another recent
oddity: Manipulating a graph in the (very nice, I repeat!) GUI dialog,
when I click "apply", sometimes the graph window content changes to a
completely different graph from literally weeks before. It's not even
related to the currently opened workfile, but something that I had
worked on earlier. (There is no dataloss, however, since I have a custom
of saving the graph as an icon to the session view before I change it.
But maybe the graph would have been lost otherwise, don't know.)
I guess there must be some gretl/gnuplot internal temp file still
lurking around somewhere which is getting activated. I think this
happened to me both on Windows and Linux.
thanks,
sven
5:35 p.m.
Sven Schreiber schrieb:
sorry for the email mini-flood, I was just reminded of another recent
oddity: Manipulating a graph in the (very nice, I repeat!) GUI dialog,
when I click "apply", sometimes the graph window content changes to a
completely different graph from literally weeks before. It's not even
related to the currently opened workfile, but something that I had
worked on earlier. (There is no dataloss, however, since I have a custom
of saving the graph as an icon to the session view before I change it.
But maybe the graph would have been lost otherwise, don't know.)
I guess there must be some gretl/gnuplot internal temp file still
lurking around somewhere which is getting activated. I think this
happened to me both on Windows and Linux.
this really happens to me often, so I will soon enter it into the bug
tracker in order to remember it.
-sven
6:44 p.m.
On Wed, 24 Feb 2010, Sven Schreiber wrote:
Sven Schreiber schrieb:
>
> sorry for the email mini-flood, I was just reminded of another recent
> oddity: Manipulating a graph in the (very nice, I repeat!) GUI dialog,
> when I click "apply", sometimes the graph window content changes to a
> completely different graph from literally weeks before. It's not even
> related to the currently opened workfile, but something that I had
> worked on earlier. (There is no dataloss, however, since I have a custom
> of saving the graph as an icon to the session view before I change it.
> But maybe the graph would have been lost otherwise, don't know.)
>
> I guess there must be some gretl/gnuplot internal temp file still
> lurking around somewhere which is getting activated. I think this
> happened to me both on Windows and Linux.
this really happens to me often, so I will soon enter it into the bug
tracker in order to remember it.
Weird, I've never seen that. However, it shouldn't be difficult to
ensure that it doesn't happen. I'll take a look at it.
Allin.
10:14 a.m.
Allin Cottrell schrieb:
On Wed, 24 Feb 2010, Sven Schreiber wrote:
> Sven Schreiber schrieb:
>
>> sorry for the email mini-flood, I was just reminded of another recent
>> oddity: Manipulating a graph in the (very nice, I repeat!) GUI dialog,
>> when I click "apply", sometimes the graph window content changes to a
>> completely different graph from literally weeks before. It's not even
>> related to the currently opened workfile, but something that I had
>> worked on earlier. (There is no dataloss, however, since I have a custom
>> of saving the graph as an icon to the session view before I change it.
>> But maybe the graph would have been lost otherwise, don't know.)
>>
>> I guess there must be some gretl/gnuplot internal temp file still
>> lurking around somewhere which is getting activated. I think this
>> happened to me both on Windows and Linux.
> this really happens to me often, so I will soon enter it into the bug
> tracker in order to remember it.
Weird, I've never seen that. However, it shouldn't be difficult to
ensure that it doesn't happen. I'll take a look at it.
hm, well if nobody else sees this maybe it's a corrupted install due to
various versions of gretl and gnuplot installed on top of each other...
thanks,
sven
1:59 a.m.
On Mon, 1 Mar 2010, Sven Schreiber wrote:
Allin Cottrell schrieb:
> On Wed, 24 Feb 2010, Sven Schreiber wrote:
>
>> Sven Schreiber schrieb:
>>
>>> sorry for the email mini-flood, I was just reminded of another recent
>>> oddity: Manipulating a graph in the (very nice, I repeat!) GUI dialog,
>>> when I click "apply", sometimes the graph window content changes
to a
>>> completely different graph from literally weeks before. It's not even
>>> related to the currently opened workfile, but something that I had
>>> worked on earlier. (There is no dataloss, however, since I have a custom
>>> of saving the graph as an icon to the session view before I change it.
>>> But maybe the graph would have been lost otherwise, don't know.)
>>>
>>> I guess there must be some gretl/gnuplot internal temp file still
>>> lurking around somewhere which is getting activated. I think this
>>> happened to me both on Windows and Linux.
>> this really happens to me often, so I will soon enter it into the bug
>> tracker in order to remember it.
>
> Weird, I've never seen that. However, it shouldn't be difficult to
> ensure that it doesn't happen. I'll take a look at it.
hm, well if nobody else sees this maybe it's a corrupted install due to
various versions of gretl and gnuplot installed on top of each other...
Well, I have "taken a look", and have put in place a safeguard
against getting an old, stale graph that I thought to be redundant
but may be needed in some odd cases. That's in CVS/snapshot. If
you see this problem again, please let me know.
Allin.
6:35 p.m.
On Tue, 16 Feb 2010, Sven Schreiber wrote:
it seems to me that gretl on windows is not as "snappy" as
it used to
be. I'm referring at least to 1.8.7 and the current snapshot, not sure
when I started seeing the behavior, as I work with gretl more often on
Linux.
I experience short but annoying delays when windows are opened and stuff
like that. I'm guessing the GTK toolkit is responsible -- I seem to
remember that an updated version has been used by gretl recently?
I don't have XP to test on any more, but I do test on Vista from
time to time. What I notice is that gretl is a bit "gummy" at
start-up (and gnuplot graphs take a while to come up at first),
compared to under Linux on the same machine. I assume this is an
OS thing ("security" checks?). Windows itself is very slow to
start compared to Linux. But once the program has been running for
a while the responsiveness seems pretty similar on the two
platforms.
Allin.
7:05 p.m.
I saw this myself. The first chart takes up to 3 seconds to display on a
brand new computer. From then on it's almost instantaneous.
On Tue, Feb 16, 2010 at 3:35 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Tue, 16 Feb 2010, Sven Schreiber wrote:
> it seems to me that gretl on windows is not as "snappy" as it used to
> be. I'm referring at least to 1.8.7 and the current snapshot, not sure
> when I started seeing the behavior, as I work with gretl more often on
> Linux.
>
> I experience short but annoying delays when windows are opened and stuff
> like that. I'm guessing the GTK toolkit is responsible -- I seem to
> remember that an updated version has been used by gretl recently?
I don't have XP to test on any more, but I do test on Vista from
time to time. What I notice is that gretl is a bit "gummy" at
start-up (and gnuplot graphs take a while to come up at first),
compared to under Linux on the same machine. I assume this is an
OS thing ("security" checks?). Windows itself is very slow to
start compared to Linux. But once the program has been running for
a while the responsiveness seems pretty similar on the two
platforms.
Allin.
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
7:32 p.m.
More like 10 seconds.
2010/2/16 Patricio Cuarón <patriciocuaron(a)gmail.com>
I saw this myself. The first chart takes up to 3 seconds to display
on a
brand new computer. From then on it's almost instantaneous.
On Tue, Feb 16, 2010 at 3:35 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
>
> On Tue, 16 Feb 2010, Sven Schreiber wrote:
>
> > it seems to me that gretl on windows is not as "snappy" as it used to
> > be. I'm referring at least to 1.8.7 and the current snapshot, not sure
> > when I started seeing the behavior, as I work with gretl more often on
> > Linux.
> >
> > I experience short but annoying delays when windows are opened and stuff
> > like that. I'm guessing the GTK toolkit is responsible -- I seem to
> > remember that an updated version has been used by gretl recently?
>
> I don't have XP to test on any more, but I do test on Vista from
> time to time. What I notice is that gretl is a bit "gummy" at
> start-up (and gnuplot graphs take a while to come up at first),
> compared to under Linux on the same machine. I assume this is an
> OS thing ("security" checks?). Windows itself is very slow to
> start compared to Linux. But once the program has been running for
> a while the responsiveness seems pretty similar on the two
> platforms.
>
> Allin.
> _______________________________________________
> Gretl-devel mailing list
> Gretl-devel(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-devel
>
4:11 p.m.
I did also installed 1.8.7cvs in Windows XP and notice this large delay in
starting gnuplot.
Didn't it used to leave a gnuplot console window opened?
I checked the Preferences page and there is not a field to use a different
gnuplot installation. Would this be an option? (Probably too much
customization for gretl's gnuplot).
Hélio
On Tue, Feb 16, 2010 at 7:42 PM, Sven Schreiber <svetosch(a)gmx.net> wrote:
Patricio Cuarón schrieb:
> I saw this myself. The first chart takes up to 3 seconds to display on a
> brand new computer. From then on it's almost instantaneous.
>
well it's not *only* that for me, but I agree that this one is the most
noticeable!
-sven
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
2:13 a.m.
On Wed, 17 Feb 2010, Hélio Guilherme wrote:
I did also installed 1.8.7cvs in Windows XP and notice this
large delay in starting gnuplot.
Didn't it used to leave a gnuplot console window opened?
No, gretl has never used a gnuplot console; we call gnuplot
non-interactively to create a PNG file, which we then display in a
GTK window that is under gretl's control.
I checked the Preferences page and there is not a field to use a
different gnuplot installation. Would this be an option?
Definitely not. There's nothing wrong with the build of gnuplot
that is in the gretl package for Windows, and it is optimized for
use with gretl.
The slowness of the first appearance of a gretl/gnuplot graph (and
please note, it's only the first appearance of a graph in a gretl
session) is almost surely a Windows issue.
I'd like to raise a more general question about the "sluggishness
of gretl on Windows" issue that Sven brought up. We're
econometricians, right? So we'd like to see evidence. Has anyone
tested the speed of current gretl against, say, gretl 1.8.0 or
1.7.0 -- or whatever version represents the good old days before
things supposedly slowed down -- on a current Windows
installation?
I'd be interested in the results of such a test, and if current
gretl really is appreciably slower than old gretl on the same
machine and OS then I'd be motivated to spend some time trying
to figure out why that might be. But right now I have no evidence,
and I can't think of any reason why it should be so, on the
gretl/GTK side. (On the other hand, I can think of several reasons
why MS Windows itself might have slowed down, on a given machine
-- from general corruption of the OS with age, to the effects of
successive security patches, to capture by a botnet.)
Allin.
9:29 a.m.
Allin Cottrell schrieb:
I'd like to raise a more general question about the "sluggishness
of gretl on Windows" issue that Sven brought up. We're
econometricians, right? So we'd like to see evidence. Has anyone
tested the speed of current gretl against, say, gretl 1.8.0 or
1.7.0 -- or whatever version represents the good old days before
things supposedly slowed down -- on a current Windows
installation?
First of all, I mostly use gretl on Linux, so it's not a big deal for
me. Second, I don't know how this could be tested, because it's about
responsiveness; what I noticed is sometimes a delay of let's say half a
second when opening a window. Actually, when it happens it feels a bit
like a GUI program in an interpreted language (e.g. Python), you know
not really bad, but noticeably slower. (Leave aside the gnuplot startup
thing.) Third, it's not fully deterministic, right now for example I
don't see these (small) problems.
But I sent the message because subjectively it seemed to me to be a
clearly recent behavior. However, I cannot say exactly at which version
it happened, for the above reason that I'm mostly on Linux.
I'd be interested in the results of such a test, and if current
gretl really is appreciably slower than old gretl on the same
machine and OS then I'd be motivated to spend some time trying
to figure out why that might be. But right now I have no evidence,
and I can't think of any reason why it should be so, on the
gretl/GTK side. (On the other hand, I can think of several reasons
why MS Windows itself might have slowed down, on a given machine
-- from general corruption of the OS with age, to the effects of
successive security patches, to capture by a botnet.)
I can't offer a proof, but with other applications I haven't noticed any
speed or responsiveness changes. So I doubt it's a generally screwed
windows installation, I think it must be something more subtle.
All in all, I'm not sure if it's worth digging deeper here -- so far it
seems other people only see the gnuplot startup delay, right?
thanks,
sven
5188
days inactive
5202
days old
12 comments
4 participants
participants (4)
-
Allin Cottrell -
Hélio Guilherme -
Patricio Cuarón -
Sven Schreiber