frysk meeting 2007-06-20 at 1330 UTC (9:30 US East Coast Time)

frysk meeting 2007-06-20 at 1330 UTC (9:30 US East Coast Time)

[Andrew Cagney]

(all welcome to dial in, please ping me for dial in info)

This week Kris will be discussing the Assembler and Memory windows esp. 
some 64-bit nits.  See:

--

Also, going forward I'll pick up the task of posting notes shortly after 
the meeting.

Andrew

Notes from frysk meeting 2007-06-20 at 1330 UTC (9:30 US East Coast Time)

[Andrew Cagney]

The goal of the meeting was to touch on and update the status of the 
issues identified by kris in the memory and disassembler windows.

Th discussion first worked through the memory window, and then the 
disassembler window.  Many of the problems, that involved the spin 
buttons for selecting addresses, were common to both.  The bugs were 
updated to reflect that.

http://sourceware.org/bugzilla/show_bug.cgi?id=4611
Memory Window: Spin buttons for address range have incorrect upper bound
-> currently it is hardwired into glade
-> bug, upper bound should be set dynamically from the task's ISA

http://sourceware.org/bugzilla/show_bug.cgi?id=4612
Memory Window: Spin buttons do not allow entry of specific addresses
-> there's a cursor but can't edit any thing
-> bug: should allow stuff to be pased in and edited
-> http://sourceware.org/bugzilla/show_bug.cgi?id=4672 should include 0x 
prefix when displaying addresses
-> http://sourceware.org/bugzilla/show_bug.cgi?id=4673 accept symbolic 
addresses (just send it to the expression evaluator for processing)

http://sourceware.org/bugzilla/show_bug.cgi?id=4613
Memory Window: Spin buttons do not enforce correct address limits
-> changed to: Memory Window: Spin buttons do not recover from incorrect 
address limits
Bug clarified to require that, when the user insists that they want to 
examine an invalid address or examine what appears to be valid 
(according to maps but isn't) address, the window should handle that 
gracefully by displaying - within the assembler or memory window an 
error indication

http://sourceware.org/bugzilla/show_bug.cgi?id=4614
Memory Window: There is no way to select what address range/space is 
being examined
Would like to be able to specify that the memory, or assembler range 
being examined is some offset into a memory segment.  Several idea's 
were floated, main useability concern was that there could be lots of 
segments - firefox had 320.  This could replace the "pc" information in 
the case of the memory window.

http://sourceware.org/bugzilla/show_bug.cgi?id=4615
Memory Window: The default number of bits used in the display is 32 even 
on 64-bit platforms
-> fixed by mcvet; discussion lead to new bugs
-> http://sourceware.org/bugzilla/show_bug.cgi?id=4674 ; The location 
should be incremented by "number of bits" or more exactly "word size"
-> http://sourceware.org/bugzilla/show_bug.cgi?id=4675 ; By default 
<location> <8-bit> and <word> colums should be displayed

http://sourceware.org/bugzilla/show_bug.cgi?id=4616
Memory Window: When selecting 64-bits display, the value column header 
still states it is 32-bits
-> fixed by mcvet

http://sourceware.org/bugzilla/show_bug.cgi?id=4617
Memory Window: Hexadecimal values are shown without leading zeros
-> agreed should both have leading zeros and be fixed width

http://sourceware.org/bugzilla/show_bug.cgi?id=4618
Memory Window: Changing the columns to display doesn't actually make 
them appear
-> fixed by mcvet

http://sourceware.org/bugzilla/show_bug.cgi?id=4619
Memory Window: Confusing text at top of window
-> see 4614, should instead display segment and offset

http://sourceware.org/bugzilla/show_bug.cgi?id=4620
Memory Window: Current program counter values appear editable but are not
-> fixed by mcvet

http://sourceware.org/bugzilla/show_bug.cgi?id=4621
Memory Window: Performing a right-click on the 'down' From spin button 
causes a hang
http://sourceware.org/bugzilla/show_bug.cgi?id=4622
Memory Window: Performing a right-click on the 'up' To spin button 
causes a hang
http://sourceware.org/bugzilla/show_bug.cgi?id=4626
Disassembly Window: left-click on 'down' From spin causes a hang
http://sourceware.org/bugzilla/show_bug.cgi?id=4627
Disassembly Window: left-click on 'down' To spin causes a hang
-> outch!
-> grouping with disassembler as should use common code

http://sourceware.org/bugzilla/show_bug.cgi?id=4623
Memory Window: Column based output is confusing and wastes window real 
estate
-> superseeded by bugs 4674, 4675

http://sourceware.org/bugzilla/show_bug.cgi?id=4624
Disassembly Window: Edit Columns button is greyed out and cannot be enabled
-> fixed by mcvet

http://sourceware.org/bugzilla/show_bug.cgi?id=4625
Disassembly Window: Impossible to increase either From or To addresses 
using spin buttons
-> fixed by mcvet

--

General discussion on memory vs disassembler window.  The key difference 
is that the disassembler window's increment or word-size is variable 
determined by each instruction that is disassembled.

[begs the question of can/should the memory window allow its word-size 
to be variable, or is that just too confusing :-)]

--

Discussion over should the windows be enabled; agreed that kris should 
commit his changes to enable these windows.

--

http://sourceware.org/bugzilla/show_bug.cgi?id=4562
Editing a session name causes a new session to be created
-> fixed in head, likely by sami fixing a similar issue in the monitor

Next meeting 27th June, technical discussion.