3424 JSP editing still slow Follow

My experience with build 3419 editing one JSP was that it was taking 10 seconds to respond to my typing. Now with build 3424, I can't even open that JSP because Idea just hangs. Oh well, it's back to 4.5.4 again.

FWIW, my CVS bugs appear to have also been relegated to the back burner.

Robert S. Sfeir wrote:

It still takes 2 seconds to hit return

Well, it doesn't take two seconds for me (P4 3.0GHz) - just about 0.5 to 1 second - but
this still feels too slow to develop with pleasure.

Also, I've just come back to editing some older JSP files and I have developed the strong
desire to get rid of this JSP formatting. It just messes up the files, even when not
reformatting the entire file but during normal edit operations like pressing <![CDATA[]]>, etc.
Those files just don't get properly formatted with the current engine, which is OK, but
there should be a way to turn this off and to stop it annoying the hell outta me.

Maybe this would even solve the performance problem...

Sascha

Ralph, where is your cpu snapshot and jira issue? We can't work
effectively in terms of "nothing works, I'm back to release. Period".

IK

Ralph Saunders wrote:

My experience with build 3419 editing one JSP was that it was taking 10 seconds to respond to my typing. Now with build 3424, I can't even open that JSP because Idea just hangs. Oh well, it's back to 4.5.4 again.

FWIW, my CVS bugs appear to have also been relegated to the back burner.

Hi Robert!

Robert S. Sfeir wrote:

It still takes 2 seconds to hit return
it still takes 2 second when you complete a tag (Like <spring:bind

and typing is still real jittery.

We tried to work on PowerBook and found the speed acceptable (at least
for your project). Try the latest build and if it's still slow send us
(me directly preferably) your cpu snapshots.

JetBrains, you've got to get a PowerBook and try things on there, not on
a monster dual G5. You shouldn't need processor horsepower to get
proper editing speed.

You are right. But there are objective reasons why JSP editing is slower
than for Java. I hope we'll change this with BF release but for now on
JSP's of big sizes (600+ str) enter will take ~0.7-1s on your PowerBook.

Now we're at RC2. It seems every EAP we go through the same circles...
everything else gets fixed first, then JSP/Web stuff gets looked at
last, and the community spends its time posting rant emails like this
one.

You know that this is far from reality. JSP improvement is one of the
major goals for this release and I hope you've noticed some changes in
this area. Also the only known major problem for JSP is performance
while other things are almost clean (my opinion goes from jira and
tracker submitted issues). BTW I hope you'll find it acceptable in the
next build (3426).

Personally until the 2 seconds to get a new line is fixed, I don't want
to file anymore bugs against JSPs, it's just too frustrating and time
consuming. This problem wasn't there in much earlier EAPs, and is
certainly not there in 4.5, so I'm guessing it's a problem with the IDE
itself not the JDK. If it's the JDK then why doesn't all this happen in
netbeans running under the same conditions?

We don't deny that we have problems and I hope you see that we are
trying to help so please, Robert, please, calm down and lets try to work
together on problems we have.

IK

In article <dc7j8k$h5o$1@is.intellij.net>,
"Igor Kuralenok (JetBrains)" <ik@intellij.com> wrote:

Hi Robert!

Robert S. Sfeir wrote:

It still takes 2 seconds to hit return
it still takes 2 second when you complete a tag (Like <spring:bind

and typing is still real jittery.

We tried to work on PowerBook and found the speed acceptable (at least
for your project). Try the latest build and if it's still slow send us
(me directly preferably) your cpu snapshots.

I did, if you mean 3425, and I did send snapshots to Max.

JetBrains, you've got to get a PowerBook and try things on there, not on
a monster dual G5. You shouldn't need processor horsepower to get
proper editing speed.

You are right. But there are objective reasons why JSP editing is slower
than for Java. I hope we'll change this with BF release but for now on
JSP's of big sizes (600+ str) enter will take ~0.7-1s on your PowerBook.

You've seen my jsps, they're like 200 lines tops. 1 second return is not
acceptable, I'm not sure how you deem that acceptable, if you had this
kind of performance in the java classes and you were working in those
all day, would you deem that acceptable?

Now we're at RC2. It seems every EAP we go through the same circles...
everything else gets fixed first, then JSP/Web stuff gets looked at
last, and the community spends its time posting rant emails like this
one.

You know that this is far from reality. JSP improvement is one of the
major goals for this release and I hope you've noticed some changes in
this area.

Sure I have, but the issues I mentioned above really ruin the whole
dinner.

Also the only known major problem for JSP is performance
while other things are almost clean (my opinion goes from jira and
tracker submitted issues). BTW I hope you'll find it acceptable in the
next build (3426).

I will try it. My JSPs still don't format right, I don't know what you
mean by almost clean, unless you mean in 3426.

Personally until the 2 seconds to get a new line is fixed, I don't want
to file anymore bugs against JSPs, it's just too frustrating and time
consuming. This problem wasn't there in much earlier EAPs, and is
certainly not there in 4.5, so I'm guessing it's a problem with the IDE
itself not the JDK. If it's the JDK then why doesn't all this happen in
netbeans running under the same conditions?

We don't deny that we have problems and I hope you see that we are
trying to help so please, Robert, please, calm down and lets try to work
together on problems we have.

My being fired up is because of a response I got that enter key would
not be fixed. If anything I've done everything I can to send in as many
snapshots as I can. I know the enter key is being fixed, but I'm not
seeing it improved. I did noitce code suggetions improve in speed a lot
in 3425, but as soon as I hit enter to accept the suggestion there is a
delay, and as soon as I close the tag and hit enter there is a delay.
Yes sent in cpu snapshots.

I'm not saying your efforts are gone to waist or not appreciated, but
you have to understand where the value is for a user. You can throw in
all the features you want, but the value to us is to be able to work at
our normal pace, and right now idea is not letting us do that in JSPs.
That reduces the value of anything you've done unfortunately because
it's the most glaring thing we see... and we've been seeing it for a
while, and noting that we're on RC build doesn't make us feel any better.

R

I'm not saying your efforts are gone to waist

Efforts going to waist, hmm... sounds dirty. How that works? A lap dance? Some other exotic exercise from Thailand to make sure your hips are in good shape? Oh my, now I'll the ghost image of a java develop doing a "waist effort" will stick in the back of my eyes for at least a few hours.

P.S.: offtopic, I know. Sorry, but I couldn't resist. This thread needs to lighten up, anyways :)

P.P.S: If you're from Thailand and know about such exotic aerobic exercises, share with us :)

heh nice typo huh, efforts going to waist == belly dancing :)

Beer-belly.

Sorry Marcus, that's not exactly lightening-up, is it?

Beer-belly.

Sorry Marcus, that's not exactly lightening-up, is it?

EW! Nice visual for the morning.

R

I cant take a CPU snapshot because the IDE hangs, accepting no input at all. I did get a snapshot after the IDE came back from it's nap, but it was 12 MB and I couldn't attach it to my other post (in a different thread). I did post the offending jsp and asked how I could get JetBrains the snapshot. Maxim sent me an answer this morning so I can now get you the snapshot.

As far as a Jira issue, I was following this thread for a while before I started posting my experiences. It has been around for a while and I understood that a Jira issue had already been filed (see the forum fhread titled "JSP editing performance" started on 21 July).

In article <42DFE4C9.5080604(at)nunya.net>,
Norris Shelton <nunya(at)nunya.net> wrote:

Is there a ticket for this? I'm seeing REALLY slow performance. Even
in .java files.

Yeha there is

You should press the CPU button once before you start the slow operation
and then a second time when the slow operation has finished. A CPU
snapshot is not like a stack dump, it needs a start and an end point
between which it will measure activity.

R

Thank you. I obviously misunderstood how that worked. I now have a reasonably sized snapshot that I can attach. and will attach to Maxim's request (in another thread).

Guys, how does it work for you in new build?

IK

It took about 5 seconds to open the jsp that appeared to hang, but once it was open the editor was responsive, with the exception of hitting enter which still took about a second. From my perspective, the opening time and time after hitting enter still should be addressed, but this build is usable IMO. Thanks.

Would you like new CPU snapshots for opening a jsp and hitting enter?

Ralph Saunders wrote:

It took about 5 seconds to open the jsp that appeared to hang, but once it was open the editor was responsive, with the exception of hitting enter which still took about a second. From my perspective, the opening time and time after hitting enter still should be addressed, but this build is usable IMO. Thanks.

Would you like new CPU snapshots for opening a jsp and hitting enter?

Yes, please. BTW how many lines there were in that JSP and how many
files with .tld extension do you have allover your project?

IK

In article <dcam9v$6qq$2@is.intellij.net>,
"Igor Kuralenok (JetBrains)" <ik@intellij.com> wrote:

Guys, how does it work for you in new build?

IK

Thumbs down, but you told me it was an Apple problem.

R

JSP performance problem seems to be sorted out in 3429