Re: [mat-dev] Proposal: Add option to display bytes in KB, MB, GB, or Sm

[Andrew Johnson <andrew_johnson@xxxxxxxxxx>]

API additions require the minor version
to be incremented.
http://wiki.eclipse.org/Version_Numbering#When_to_change_the_minor_segment

Normally service releases don't have
API changes.

The Memory Analyzer project generally
increments the minor version for each annual release (the project, feature
and plug-in version numbers don't have to increment in step, but we have
done that for releases so far).

I'm not sure of the best way around
this - wait until near the next release to make this available; branch
the code now so that the Luna updates are just service updates and make
changes to trunk; make the changes now in a way that doesn't add APIs then
move/change them to APIs later.

Andrew Johnson



From:      
 Kevin Grigorenko <kevin.grigorenko@xxxxxxxxxx>
To:      
 Memory Analyzer Dev
list <mat-dev@xxxxxxxxxxx>
Date:      
 01/07/2014 19:38
Subject:    
   Re: [mat-dev]
Proposal: Add option to display bytes in KB, MB, GB, or Smart formats
Sent by:    
   mat-dev-bounces@xxxxxxxxxxx

---

I'm trying to apply this patch to the
latest 1.4 code and I'm running into some API version issues. I set the
API baseline to MAT 1.4, and applied the patch. The build error is:

The minor version should be incremented in version 1.4.0, since new APIs
have been added since version 1.4.0 MANIFEST.MF /org.eclipse.mat.report/META-INF
line 6 Version Numbering Problem

I tried to change the Bundle-Version to 1.4.1.qualifier and updated the
@since tags in the new classes to 1.4.1, but it continues to complain as
it seems to want to update the version to 1.5.0.qualifier.

If I update Bundle-Version to 1.5.0.qualifier and update the @since tags
to 1.5, everything works. Is this expected? Is this what I should do?

--
Kevin Grigorenko
IBM WebSphere Foundation SWAT Team
kevin.grigorenko@xxxxxxxxxx
Blog: https://www.ibm.com/developerworks/mydeveloperworks/blogs/kevgrig/?lang=en

Kevin
Grigorenko---06/27/2014 11:04:46 AM---Hi Krum, Thanks for taking a look.
> It gets a bit more difficult to read if I have a list of entrie

From:	Kevin Grigorenko/Austin/IBM@IBMUS
To:	Memory Analyzer Dev list <mat-dev@xxxxxxxxxxx>
Date:	06/27/2014 11:04 AM
Subject:	Re: [mat-dev] Proposal: Add option to display bytes in KB, MB, GB, or Smart formats
Sent by:	mat-dev-bounces@xxxxxxxxxxx

---

Hi Krum, Thanks for taking a look.

> It gets a bit more difficult to read if I have a list of entries sorted
by size
> And I found it pretty difficult to read in views where the objects
are not sorted by size
> Therefore I would suggest to stick to normal bytes as default, and
give the users the option to change

That makes sense and I agree, I will change the default to Bytes. One other
idea would be to add a "Default" option which uses Bytes in most
places and Smart in places such as Pie chart and leak suspects - what do
you think of that as a default?

> When using “smart” the bytes do not have a unit displayed. This
may be confusing. Still if I have only bytes, I would prefer leaving the
unit away (less info). Do you think you could make it this way?

Yes, I will modify to do that.

> I think this is enough. Just let’s discuss a bit more on the couple
of remarks about the Smart numbers.

Great, thanks! I will make the modifications above and submit a new patch
and also give some time for others to provide thoughts if they'd like.

--
Kevin Grigorenko
IBM WebSphere Foundation SWAT Team
kevin.grigorenko@xxxxxxxxxx
Blog: https://www.ibm.com/developerworks/mydeveloperworks/blogs/kevgrig/?lang=en

"Tsvetkov,
Krum" ---06/27/2014 02:26:47 AM---Hi Kevin, after going quickly through
the code changes I applied them locally and played with them.

From:	"Tsvetkov, Krum" <krum.tsvetkov@xxxxxxx>
To:	Memory Analyzer Dev list <mat-dev@xxxxxxxxxxx>
Date:	06/27/2014 02:26 AM
Subject:	Re: [mat-dev] Proposal: Add option to display bytes in KB, MB, GB, or Smart formats
Sent by:	mat-dev-bounces@xxxxxxxxxxx

---

Hi Kevin,

after going quickly through the code changes I applied them locally and
played with them.

I find the change very nice and would encourage you to submit it. I still
have some remarks:

>> * The default option is BytesDisplay.Smart which changes the default
MAT display behavior (which would be BytesDisplay.Bytes).

-       I
really liked (found it easier to read) how the values are rendered with
the Smart option in the Pie chart and in the Leak suspects text, i.e. where
we have just one or a few values
-       It
gets a bit more difficult to read if I have a list of entries sorted by
size, e.g. the Dominator Tree
-       And
I found it pretty difficult to read in views where the objects are not
sorted by size, e.g. different reference chains. Having only bytes gave
me immediate overview which the big objects are (just the optical length
of the number, I don’t really have to read it). However, having a mixture
of the different sizes I need to read more or less every single one to
spot a bigger object. Here is an example:
   
Class Name              
                     
                     
                     
      | Shallow Heap | Retained Heap
--------------------------------------------------------------------------------------------------------------------------------
org.eclipse.mat.parser.internal.SnapshotImpl @ 0xc230d258
                     
                 |  
        64 |       7,11 MB
|- source org.eclipse.mat.parser.model.ClassImpl @ 0xc02fbe60
                     
             |      
   104 |           248
|  '- data org.eclipse.swt.widgets.TreeItem @ 0xc31b55c8
                     
                  |  
        88 |           112
|     '- [3] org.eclipse.swt.widgets.TreeItem[4] @ 0xc31b90e0
                     
             |      
    32 |           120
|        '- items org.eclipse.swt.widgets.Tree @ 0xc2107e30
                     
               |    
     288 |       1,88 KB
|           |- [58], [59], [60] org.eclipse.swt.widgets.Control[1024]
@ 0xc22787d8
              |      4,02
KB |       6,69 KB
|           |  '- controlTable org.eclipse.swt.widgets.Display
@ 0xc1e21e60
Native Stack         |        
 624 |     146,71 KB
|           |- control org.eclipse.swt.accessibility.Accessible
@ 0xc21083d0
                    |
         160 |       1,33 KB
|           '- Total: 2 entries    
                     
                     
                 |  
           |        
     
|- snapshot org.eclipse.mat.ui.snapshot.editor.HeapEditor$SnapshotEditorInput
@ 0xc204e980
      |           32 |    
      112
|- snapshot org.eclipse.mat.internal.snapshot.inspections.DominatorQuery$DefaultTree
@ 0xc2cc8388|
          40 |          
 40
|- snapshot org.eclipse.mat.snapshot.query.ObjectListResult$Outbound @
0xc31774c0
               |    
      24 |            24
--------------------------------------------------------------------------------------------------------------------------------
   
-       Therefore
I would suggest to stick to normal bytes as default, and give the users
the option to change. This is just my opinion, I’d be happy to hear some
other comments on this.
-       When
using “smart” the bytes do not have a unit displayed. This may be confusing.
Still if I have only bytes, I would prefer leaving the unit away (less
info). Do you think you could make it this way?
 
>> * The additional instances of the Bytes class might add some performance
overhead.

I think these are usually only for the displayed records, which should
be fine.

>> * I tried to make the most surgical change possible, which means
I did not audit all the code for all uses of bytes, but only particularly
the structured results and reports that used a NumberFormat or DecimalFormat
and switched those to ByteFormat (where applicable).

I think this is fine. If we spot afterwards some places which also need
to be adapted we could fix them.

>> * All JUnit tests passed - is there anything else that needs to
be done before committing (in addition to any feedback)?

I think this is enough. Just let’s discuss a bit more on the couple of
remarks about the Smart numbers.

Krum

From: mat-dev-bounces@xxxxxxxxxxx [mailto:mat-dev-bounces@xxxxxxxxxxx]
On Behalf Of Kevin Grigorenko
Sent: Donnerstag, 26. Juni 2014 15:57
To: Memory Analyzer Dev list
Subject: Re: [mat-dev] Proposal: Add option to display bytes in KB,
MB, GB, or Smart formats

Thanks! Take your time

--
Kevin Grigorenko
IBM WebSphere Foundation SWAT Team
kevin.grigorenko@xxxxxxxxxx
Blog: https://www.ibm.com/developerworks/mydeveloperworks/blogs/kevgrig/?lang=en


"Tsvetkov, Krum" ---06/26/2014 05:30:03 AM---Hi Kevin, Sorry for the late reply.  

From:	"Tsvetkov, Krum" <krum.tsvetkov@xxxxxxx>
To:	Memory Analyzer Dev list <mat-dev@xxxxxxxxxxx>
Date:	06/26/2014 05:30 AM
Subject:	Re: [mat-dev] Proposal: Add option to display bytes in KB, MB, GB, or Smart formats
Sent by:	mat-dev-bounces@xxxxxxxxxxx

---

Hi Kevin,

Sorry for the late reply.
I like the idea to offer the possibility to display the sizes in different
units.

I am currently looking into the patch. I may need some more time as it
is pretty big. I hope afterwards I could answer your questions/concerns.

Krum


From: mat-dev-bounces@xxxxxxxxxxx
[mailto:mat-dev-bounces@xxxxxxxxxxx]
On Behalf Of Kevin Grigorenko
Sent: Dienstag, 17. Juni 2014 22:35
To: mat-dev@xxxxxxxxxxx
Subject: [mat-dev] Proposal: Add option to display bytes in KB, MB,
GB, or Smart formats

Proposed patch to display bytes in KB, MB, GB, or Smart formats: https://bugs.eclipse.org/bugs/show_bug.cgi?id=437626 Diff: https://bugs.eclipse.org/bugs/attachment.cgi?id=244315&action="">

A few specific things I wanted to raise on the dev list:
* The Bytes class is added as a logical wrapper around a number of bytes. The Column class maps this to the BytesFormat formatter.
* The default option is BytesDisplay.Smart which changes the default MAT display behavior (which would be BytesDisplay.Bytes).
* The additional instances of the Bytes class might add some performance overhead.
* I tried to make the most surgical change possible, which means I did not audit all the code for all uses of bytes, but only particularly the structured results and reports that used a NumberFormat or DecimalFormat and switched those to ByteFormat (where applicable).
* All JUnit tests passed - is there anything else that needs to be done before committing (in addition to any feedback)?

--
Kevin Grigorenko
IBM WebSphere Foundation SWAT Team
kevin.grigorenko@xxxxxxxxxx
Blog: https://www.ibm.com/developerworks/mydeveloperworks/blogs/kevgrig/?lang=en_______________________________________________
mat-dev mailing list
mat-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mat-dev_______________________________________________
mat-dev mailing list
mat-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mat-dev_______________________________________________
mat-dev mailing list
mat-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mat-dev_______________________________________________
mat-dev mailing list
mat-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/mat-dev

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU