Demetra - code-name for what version?
What version is this the code-name for?
I've just been informed that some fairly critical bugs regarding J2EE deployment will be fixed in the final version of Demetra, and I am under the rather worrying impression that this was the code-name for version 6.0.
请先登录再写评论。
Hello Simon,
SK> What version is this the code-name for?
SK>
SK> I've just been informed that some fairly critical bugs regarding
SK> J2EE deployment will be fixed in the final version of Demetra, and I
SK> am under the rather worrying impression that this was the code-name
SK> for version 6.0.
That impression is true.
--
Dmitry Jemerov
Software Developer
JetBrains, Inc.
http://www.jetbrains.com
"Develop with pleasure!"
In which case I'm pretty bloody annoyed.
I've only just convinced my company to fork out money for buying licenses for IDEA, with one of the main features being its J2EE deployment functionality.
Being honest, the release of version 5.0 has been well below standard. We've had major issues due to CVS merging issues and the CVS integration still seems flaky. Performance issues continue to plague the EAP builds and made the initial release unusable on many of our developers machines.
To then be informed that bugs that have existed since IDEA 4.5 aren't going to be fixed until the next paid version, bugs which currently make the J2EE deployments unusable, just makes me angry.
Hello Simon,
Could you give some details about performance issues you faced? Could it be possible to obtain a CPU snapshot illustrating the
problem - that would really help us to fix them.
--
Best regards,
Eugene Zhuravlev
Software Developer
JetBrains Inc.
http://www.jetbrains.com
"Develop with pleasure!"
Oh my.
Idea 5.0 is really below previous standards and worse than 4.5 in many
areas.
Now you are telling me that
a) I have to wait a year or so, to see improvements.
b) I have to pay for improvements.
c) Those improvements will probably not be in Idea's once-upon-a-time
strengths (code editing, refactoring).
For us it might well be "too late, too little" to keep using Idea.
Currently part of the team already uses JDeveloper (for better J2EE
support) and the other part is still using Idea, but complaining about
5.0 consistently.
So the first half will probably be reluctant to switch back to Idea
when 6.0 is out and the other half will be looking for better
alternatives for a year - and maybe switching.
I had really hoped for a 5.1 that fixes some of the editing and
usability bugs and introduces at least some support for J2EE standards.
Anyway, I wish you luck...
For the records: I have recently filed about a dozen or so Jira issues
for pretty obvious bugs. Close to zero response from JetBrains yet.
Many older - even more obvious bugs - are kind of buried alive in Jira.
In fact I stopped reporting many bugs, because I see no sense to it.
Dmitry Jemerov (JetBrains) wrote:
We still suffer with poor CVS performance and J2EE module building is still a problem as well. I'll try and get some CPU profiles for you.
General editor performance has generally improved over the last few EAP builds, but unless you know about the "magic" JVM settings for matching the JVM with the OS thread priorities, it's still sluggish. Even with the JVM settings corrected, it still feels at times like working with a slightly drunk IDE - it gets there, but there's a continual feeling of slight lag. It's one of the main problems I used to have with Eclipse, which is why I changed IDE.
To be honest the performance doesn't bother me as much as this bugfix release push-back, to what I consider core functionality. I wouldn't mind as much if they were minor inconveniences, but they are critical bugs which mean that we can only deploy our J2EE applications reliably using Ant.
http://www.jetbrains.net/jira/browse/IDEADEV-2697
Simon and I have both posted CPU snapshots there. Please give us some feedback and let us help you track down the problem ASAP.
Tim
Please do not be mistaken by the "version to fix" attribute. The problem is being analysed now and if it is possible to fix it in
update release we'll certainly do.
--
Best regards,
Eugene Zhuravlev
Software Developer
JetBrains Inc.
http://www.jetbrains.com
"Develop with pleasure!"
I have the same impression as Stephen. The number of people who might otherwise be making quality contributions in the EAP has been dropping and consequently the quality of IDEA has been dropping steadily on performance, bugs, usability, well-thought-out-new-features, and "develop with pleasure". Why bother to post under these circumstances when the feedback is either ignored or not comprehended the vast majority of the time compared to previous EAP cycles. Sure, one can point out an exception or two like Inspections, but Dave G. & company are mortals and can only do so much with their code contributions and feedback. And, no, Hani, I'm not talking about feedback from rabid mobs and the unwashed masses who want IDEA to directly support every pet framework.
IDEA is in crisis despite the repeatedly expressed sincere intentions of the JetBrains employees to work with feedback. Management/leadership at Jetbrains needs to confront the reality here and return IDEA on a path to a promising future.
Hello Stephen,
1. Would you please be a bit more specific? Particulary, what fixes exactly
have been marked as done (or planned to) in Demetra, which you think you
need earlier?
2. What areas 4.5 is better than 5.0? Please refrain from general words like
"performance", "usability", "stability" or whatsoever. Any concrete things
are highly appreciated.
And yes, 5.1 is indeed planned. It doesn't have a codename since we do not
think it is necessary. As to what exactly will be done there that's another
story.
Thanks.
-
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"
Hello Jon,
Well, the feedback processing quality is somewhat proportional to the overall
quality of the feedback itself and counter-proportional to its volume. Now
think where crisis (if any) comes from.
Friendly,
-
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"
I think the reverse is true...a sustained low level of response to quality feedback & final releases rife with problems lowers interest in quality feedback and leaves you with an even lower proportion of quality feedback. This is in comparison to older EAP's such as pre-IDEA 5.x, 4.5. It is a viscous cycle.
Sure, you still have a tiny handful of dedicated people giving quality feedback who are tolerant of the lower level of responsiveness, but in the past IDEA had a far broader base of quality contributors when it could be said "Give Feedback With Pleasure" because a serious effort would almost always be make to understand the feedback and then serious consideration given to incorporating the feedback. Now feedback seems more like a shot in the dark despite the special occassions like now when you are personally paying very close attention.
Here's a concrete problem from my point of view.
I've hit the following exception in every build from early in the last EAP until the current (3526).
ava.lang.NullPointerException
at com.intellij.codeInsight.daemon.impl.GeneralHighlightingPass$1.run(GeneralHighlightingPass.java:8)
at com.intellij.psi.impl.PsiManagerImpl.performActionWithFormatterDisabled(PsiManagerImpl.java:374)
at com.intellij.codeInsight.daemon.impl.GeneralHighlightingPass.a(GeneralHighlightingPass.java:126)
at com.intellij.codeInsight.daemon.impl.GeneralHighlightingPass.doCollectInformation(GeneralHighlightingPass.java:48)
at com.intellij.codeInsight.daemon.impl.TextEditorHighlightingPass.collectInformation(TextEditorHighlightingPass.java:6)
at com.intellij.codeInsight.daemon.impl.UpdateThread$2.run(UpdateThread.java:11)
at com.intellij.openapi.application.impl.ApplicationImpl.runReadAction(ApplicationImpl.java:399)
at com.intellij.codeInsight.daemon.impl.UpdateThread.a(UpdateThread.java:28)
at com.intellij.codeInsight.daemon.impl.UpdateThread.access$100(UpdateThread.java:1)
at com.intellijj.codeInsight.daemon.impl.UpdateThread$1.run(UpdateThread.java:0)
at com.intellij.openapi.progress.impl.ProgressManagerImpl.runProcess(ProgressManagerImpl.java:77)
at com.intellij.codeInsight.daemon.impl.UpdateThread.run(UpdateThread.java:38)
I submit by 'blame core' and see entries in the tracking system. It never gets commented on, and it never gets fixed.
It happens more frequently when I do some minor edits in the CVS diff window brought up through the commit project dialog. But it will happen randomly any time I do the diff, even without edits.
I've given votes to some builds versions, but I don't really feel like I should have to shift them forward with every build I try.
Please fix this bug.
Undoubtely true. Now if one knew the way to break this cycle...
Two schemes come into my mind: hire a team of feedback screeners (bad idea
actually, direct developers communication had always been a major point in
EAP) or limited, invitation based EAP.
What do you think?
-
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"
Quite strange thing, kinda "impossible"... What plugins do you have installed?
What kind of file you've been editing in that CVS diff window? xml, java?
-
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"
I really don't know the ideal solution. Neither proposal feels quite right; you don't want to hijack direct communication with developers, and you don't want to prevent new quality contributors from coming on board. Maybe trusted volunteer screeners who become advocates for issues they back? For the part of the problem that is clarity of communication, if the volunteer screener can come to comprehend & advocate the issue, then they can help clarify and communicate the issue for the IDEA-developer audience as well as greatly improve the signal-to-noise ratio. This benefits would be:
1. Greatly reduce the number of dead-end issues that have to be examined by developers
2. Improve the quality of the issues finally submitted to developers due to pre-clarification by screener
3. Give JetBrains developers a much higher degree of faith in the quality of issues that do make it to their desk due to #3 & #1 so that they don't inadvertantly dismiss a quality issue.
Jon
I'm editing a plain java file.
My plugins directory:
InspectionGadgets
IntentionPowerPack
JavascriptLanguage
JsHtmlBridge.jar
PerforceIntegration
cvsIntegration
devkit
images
j2meplugin
jsr45debug
starteamIntegration
svn4idea
tomcat
vssIntegration
weblogicIntegration
zkm
Maxim Shafirov wrote:
I'm regularly getting the same exception during CVS diff, which has been often reported,
but never been commented on. Happens with both XML (Ant) files as well as Java code. I'm
pretty sure it's not a plugin issue, but just in case, how could a plugin cause this?
I suspect it's a concurrency problem, maybe caused by the fact that I have configured the
"Compare with latest version" CVS command into the toolbar with the customization mechanism.
Sascha
Just noticed that split between addon plugins and base.
.. 7/26/2005 11:02 Copyright 7/26/2005 11:02 ideajad 7/26/2005 11:02 RegexPlugin 7/26/2005 11:02 ]]> SvgViewer7/26/2005 11:02 . 7/26/2005 11:02
7/26/2005 10:56 198,548 tabifier-5.1.2.jar
I honestly believe that your existing EAP process is one of the best I've been part of, due primarily to the direct communication with developers. The openness of the EAP is also a bonus I believe.
A couple of suggestions which I believe would improve the existing EAP process would be:
- Turn on the CPU and memory profiler options automatically for EAP builds, so that people who don't frequent the boards know that these are available for reporting issues without having to add JVM properties.
- On an EAP release give an overview of exactly what bugs/issues have been fixed in the release notes, as the communication of "Bug fixes" is a bit vague at the moment, although I'm not suggesting mentioning specific JIRA entries.
- On a similar note to the last one; if, for example, you know that a few regular people have mentioned issues with performance in certain areas, mention it in the EAP build notes so that the greater population can also give you feedback and help you to track down issues faster.
- Ask people to give you a profile of exactly what system they are using for the EAP, so that you can narrow down issues to particular hardware/software configurations. I know JIRA asks you for the environment you are working in, but entered data tends to be very vague. I only mention this as we've discovered that a specific virus checker caused huge problems with IDEA, and we've also faced significant issues with performance on non-hyperthreaded CPUs which haven't been noticed at all on HT-enabled CPUs.
Hello Maxim,
Whether we like it or not, the communication adds overheads to the whole
development process. From what it appeared during the past two-three EAP
cycles, your resources were kind of maxed out, unable to keep both the communication
and the development quality at the levels this community was used to see
them. So one of them was "sacrificed", in order to keep the other one floating.
It seems your inclination is to, somehow, minimize the volume of (bad) data
the developers have to deal with, in order to make it easier for them to
cope with it. How about going the other way around, and make it so they can
deal with all of it? While adding resources to a team is not always a good
idea (and you're in a better position to appreciate your optimal team size),
why don't you try (again) something that was suggested several times on these
forums in the past couple of years: smaller, shorter cycles, with fewer new
features and more attention to detail.
AFAIR that's how the old EAPs (yeah, the ones everybody reffers to when they
complain :) used to be. If my memory serves me well the EAP for 3.0 was about
half a year and, man, was that (release) a breakthrough!
Just my 2c,
Andrei
>> I think the reverse is true
>>
MS> Undoubtely true. Now if one knew the way to break this cycle...
MS>
MS> Two schemes come into my mind: hire a team of feedback screeners
MS> (bad idea actually, direct developers communication had always been
MS> a major point in EAP) or limited, invitation based EAP.
MS>
MS> What do you think?
MS>
MS> -
MS> Maxim Shafirov
MS> JetBrains, Inc
MS> http://www.jetbrains.com
MS> "Develop with pleasure!"
Some sort of QA team? Like Mozilla QA.
"Andrei Oprea" <andrei.oprea@rogers.com> wrote in message
several times on these
cycles, with fewer new
Smaller cycles correlate to more detached and modular
structure, so the portion of the code could be verified and
updated by one team member, while another would work on a
different module. Then they would connect as black boxes.
IDEA does not seem as modular as say Eclipse right now. So,
when IDEA got fatter, it cannot bear the modifications. One
change can affect the whole application.
(Above is all off my imagination, I did not see code of
IDEA).
reffers to when they
EAP for 3.0 was about
Still am using it. Was reluctant to upgrade to 4.x. Was even
more unsure to get 5.x, and so I did not. I am considering
getting JFormDesigner now. With this duo I should be better
off for both server-side java stuff, as well as for Swing
dev. I have not switched to 1.5 now, so I can live without
support for annotations or generics.
I stopped using EAPs after JetBrains switched to Jira. Could
not get used to it.
Michael.
Hello Maxim,
I would very much like to give a detailed, comprehensive answer.
Unfortunately I can't because I am working long and hard hours
to finish a project.
So I can give only some example.
I always felt Idea was very stable. In Irida I set the auto save period
to ten seconds, so that I don't loose any code changes, if I have to
kill the process. I reported only the the first of these crashes:
IDEA-5393
No comment from JetBrains yet. I strongly feel that these kind of bugs
should be highest priority and should be addressed in a bug fix release
ASAP. I even isolated the problem and gave exact source code and
description to reproduce it.
I refrained from filing more issues like these as they are time consuming
to test and reproduce and I have better things to do do than speaking
into the void.
I guess for each of these crashes that get reported you have dozens of
potential customers that just won't buy Idea after evaluation...
Then formatting in general is worse than before. Sometimes it even
breaks my code: IDEA-5415
Sometimes it just renders it illegible:
Before:
@Embedded
@AttributeOverrides({
@AttributeOverride(name="_intCode", column=@Column(name="computed_destination_intcode")),
@AttributeOverride(name="_countryCode", column=@Column(name="computed_destination_countrycode", insertable=false, updatable=false)),
@AttributeOverride(name="_areaCode", column=@Column(name="computed_destination_areacode", insertable=false, updatable=false)),
@AttributeOverride(name="_subscriber", column=@Column(name="computed_destination_subscriber", insertable=false, updatable=false)),
@AttributeOverride(name="_extension", column=@Column(name="computed_destination_extension", insertable=false, updatable=false)),
@AttributeOverride(name="_complete", column=@Column(name="computed_destination_telno", insertable=false, updatable=false))
})
private TelNoImpl _computedDestination;
After:
@Embedded
@AttributeOverrides({@AttributeOverride(name = "_intCode", column = @Column(
name = "computed_destination_intcode")), @AttributeOverride(name = "_countryCode", column = @Column(
name = "computed_destination_countrycode", insertable = false, updatable = false)), @AttributeOverride(
name = "_areaCode", column = @Column(name = "computed_destination_areacode", insertable = false,
updatable = false)), @AttributeOverride(name = "_subscriber",
column = @Column(
name = "computed_destination_subscriber",
insertable = false,
updatable = false)), @AttributeOverride(
name = "_extension", column = @Column(name = "computed_destination_extension", insertable = false,
updatable = false)), @AttributeOverride(name = "_complete",
column = @Column(
name = "computed_destination_telno",
insertable = false,
updatable = false))})
private TelNoImpl _computedDestination;
There are a number of other (minor) issues, that just take out the pleasure.
In 4.5 we had the policy of always having "reformat before commit" checked.
In Irida we found it too dangerous and creating too much CVS noise.
Inspections in Idea are not reliable, which is worse than having no inspections
at all: IDEA-5399
Regardless of the tab closing policy I selected, Idea keeps closing tabs I worked
with frequently and let others stay open indefinitely. Recent-File list is also
wrong. I don't remember the JIRA issue (was from somebody else).
This is very annoying if you are used to Ctrl-E + Enter to jump back and forth
between two files. Now I often stare on the tab headers for some seconds, then
fail to find the file I want and have to do ctrl-n and some typing to open it
again.
I had troubles getting CVS to work: IDEA-2978
This was working fine before.
Idea forgets settings I made (after restart): IDEA-3482
The Run/Debug console no longer scrolls to the bottom: IDEA-5470
Intention "Remove Parameter" is now first in list (wasn't before), so that
frequently I remove a parameter I just entered manually when I meant to
create a field for it.
Remember that these are just examples. If I think about it some more,
I can come up with many more.
The overall impression is what sticks: Irida fails to meet the
"Develop with Pleasure" claim.
I fondly remember the times when I frequently thought "gosh - how did they do
this" and discovering new well-thought out and nicely implemented features
(e.g. I typed Ctrl-W once to often and instinctively tried Ctrl-Shift-W
- it just worked).
Maxim Shafirov wrote:
>> I had really hoped for a 5.1
>> Anyway, I wish you luck...
Full agreement.
The best on your comment is to show the disappointment that the
IDEA-feeling has lost somehow. EAPler are getting somewhat silent
these days :(
Stephen Kelvin schrieb:
>> Hello Stephen,
>>
>> 1. Would you please be a bit more specific? Particulary, what fixes
>> exactly have been marked as done (or planned to) in Demetra, which you
>> think you need earlier?
>> 2. What areas 4.5 is better than 5.0? Please refrain from general
>> words like "performance", "usability", "stability" or whatsoever. Any
>> concrete things are highly appreciated.
>>
>>> I had really hoped for a 5.1
>>
>>
>> And yes, 5.1 is indeed planned. It doesn't have a codename since we do
>> not think it is necessary. As to what exactly will be done there
>> that's another story.
>>
>>> Anyway, I wish you luck...
>>
>>
>> Thanks.
>>
>> -
>> Maxim Shafirov
>> JetBrains, Inc
>> http://www.jetbrains.com
>> "Develop with pleasure!"
>>
>>
Stephen Kelvin wrote:
>
>
True and false. It looks like some of the settings are lost when an EAP
is installed. Makes it look like things are broken. Just as unsettling
and takes time to figure out what is missing.
>
>
>
Ditto.
>
>
I feel your pain. I asked around in the forums, but no one else said
they had this problem. I thought I filed a ticket, but I can't find it.
>
I have had some weird CVS problems. I also use TortoiseCVS. I could
not figure out which one caused the problems.
>
This only happened to me on one thing. It was a plugin setting. I
talked with the author and added the setting manually and everything
worked great.
>
I use JBoss and this happens sometimes. I have to put my cursor on the
last line and hit CTRL-end to keep it down there. This may be by design
to allow you to time to look at something. Maybe a pin moniker would be
better. Always show last line unless pinned?
>
>
>
>
>> Hello Stephen,
>>
>> 1. Would you please be a bit more specific? Particulary, what fixes
>> exactly have been marked as done (or planned to) in Demetra, which
>> you think you need earlier?
>> 2. What areas 4.5 is better than 5.0? Please refrain from general
>> words like "performance", "usability", "stability" or whatsoever. Any
>> concrete things are highly appreciated.
>>
>>> I had really hoped for a 5.1
>>
>> And yes, 5.1 is indeed planned. It doesn't have a codename since we
>> do not think it is necessary. As to what exactly will be done there
>> that's another story.
>>
>>> Anyway, I wish you luck...
>>
>> Thanks.
>>
>> -
>> Maxim Shafirov
>> JetBrains, Inc
>> http://www.jetbrains.com
>> "Develop with pleasure!"
>>
>>
>
Hello Michael,
MJ> "Andrei Oprea" <andrei.oprea@rogers.com> wrote in message
MJ>
>> why don't you try (again) something that was suggested
>>
MJ> several times on these
MJ>
>> forums in the past couple of years: smaller, shorter
>>
MJ> cycles, with fewer new
MJ>
>> features and more attention to detail.
>>
MJ> Smaller cycles correlate to more detached and modular structure, so
MJ> the portion of the code could be verified and updated by one team
MJ> member, while another would work on a different module. Then they
MJ> would connect as black boxes. IDEA does not seem as modular as say
MJ> Eclipse right now. So, when IDEA got fatter, it cannot bear the
MJ> modifications. One change can affect the whole application.
Regardless of how the app is written, if you apply some of the rules of agile
development it doesn't matter how many people are working together and on
what, you can still work in 30 day sprints with a set of features to implement
in those 30 days. During the 30 days you can have 2 iterations to push out
to EAP so people can test what you have so far and report bugs. At the end
of the sprint you should have the features you thought you could implement
done. If not what ever is not done gets pushed to the next sprint, and something
in the next sprint gets pushed back too since you over estimated how long
it would take to write a feature. As time goes by you will really know how
long something takes to implement.
You already have a backlog to work on, all you have to do is move the features
for the sprint from the backlog to the iteration, so part of your process
is there, it's a matter of setting a goal for each sprint and push out only
what is in that sprint. If a new feature request comes in (user story) just
put it in the backlog and before the next sprint review your plan and put
together a list of things for it.
Works for many projects I've worked on, delivers value to the customer over
a period of time, and most of the time (not all the time) you end up with
a product that is more bug free and delivered on time.
jmho. Not shoving the concept down anyone's throat.
R
Hello Stephen,
IDEA-5393 - bug, will be fixed in 5.0.3
IDEA-5415 - bug, will be fixed in 5.0.3
Annotation formatting: Are those annotations public? Can I get those to play
with? Anyway that's not a bug but missing feature, which haven't been requested
during 5.0 EAP.
IDEA-5399 - known problem. Fix will require public API to change, which we
can't afford in bugfix update.
IDEA-2978 - further investigation necessary.
IDEA-3482 - not repeatable. Works fine.
IDEA-5470 - not a bug. Works the same way it worked since version 3.0
Recent files - hardly reproducible yet known problem. Will be fixed in bugfix
update.
Intention order - new feature request.
Well, it took me 1.5 hours to get through these 9 with 2 false alarms. And
we usually have 10 - 50 new JIRA items a day not to mention newsgroup posts.
And the average true/false alarm is much worse most of the time.
-
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"
Update: I was able to reproduce IDEA-3482. Thanks Stephen, this will be fixed
in 5.0.3
-
Maxim Shafirov
JetBrains, Inc
http://www.jetbrains.com
"Develop with pleasure!"
Maxim Shafirov (JetBrains) wrote:
I can reproduce this using the described steps. Change the option from the intention's
popup menu!
I think I've seen this as well. The console does scroll, but it's not always showing the
very bottom of the output. I also suspect this is something with the amount and/or the
speed the prints are made by the application or some problem calculating the current
scrollbar position - maybe L&F dependent?
Umm, no. This has been agreed on during the EAP and has been implemented in about 3 or 4
EAP builds before it reverted back to the old behavior:
http://www.intellij.net/forums/thread.jsp?forum=22&thread=128147&message=3524945
Sascha
Are you saying that because it looks impossible you wont attempt to fix it? Do you need any more information to help diagnose?