Fragile release 2.6.3.216

12 views
Skip to first unread message

C Burch

unread,
Mar 2, 2011, 1:14:59 PM3/2/11
to log...@googlegroups.com
I've just uploaded yet another fragile release. This is still meant as
a "final release." I still have about 4 minor bugs to address, and I'm
sure there will be further fragile releases. But my list of what
remains for 2.7.0 doesn't include anything that would affect the
application's visual appearance, its translations, or its
documentation. I have begun redoing screenshots for the English
documentation, and it should be safe for other translators to do
likewise.

The changes in this release:

* The icon had not yet been amended to match the modified new appearance.

* When you changed one of the bit attributes on a splitter, it did not
immediately have an effect on the computation of width conflicts.

* Right-clicking a splitter brings up a pop-up menu with "Distribute
Ascending" and "Distribute Descending" for reordering the bits more
easily than editing each individual item. This isn't quite what Ilia
suggested, but it is easier and less bug-prone to implement.

(Ilia also suggested that changing the Appearance of a splitter from
left-handed to right-handed should preserve the locations of the split
ends and place the combined end elsewhere. I agree this would be
desirable, but it is complicated (and bug-prone) to accomplish within
Logisim's current architecture, where attributes are changed in an
object that isn't really aware of the splitter that it effects - and
where the underlying assumption is that component's locations don't
change once they are placed on the canvas. I envision changing the
architecture later, but that's a huge change. Anyway, I've noted it as
something that should be addressed in a later version.)

-Carl

Ilia Lilov

unread,
Mar 2, 2011, 2:26:53 PM3/2/11
to log...@googlegroups.com
"Distribute Ascending" and "Distribute Descending" features have 3 bugs now:
1) These pop-up menu items are available only when one or more "Bit x"
attributes are changed manually, although at least one of these items
has to be available any time (bits can not be distributed ascending
and descending at the same time);
2) Both pop-up menu items distribute bits ascending;
3) Names of these pop-up menu items don't use strings from
circuit.properties now (even English GUI is chosen).

I'm going to start remaking screen shots tomorrow.

Ilia.

C Burch

unread,
Mar 2, 2011, 10:32:13 PM3/2/11
to log...@googlegroups.com
Thanks, Ilia, for the notes. The problems here are a bit embarrassing.
I checked the splitter for the case when the number of bits exceeds
Fan Out (which is the harder case), and it works for that; but I
neglected to check the easy case, when the number of bits is less than
or equal to Fan Out. That was also the cause of the problem with
enabling menu items.

I also repaired the issue with using strings. (The problem there was
forgetting a Subversion update for the directory from which I build
release versions, so the created fragile release didn't match what I
had checked into Subversion.)

Anyway, all these have been repaired, though I don't think they merit
another fragile release in themselves. Thanks again.

-Carl

> --
> You received this message because you are subscribed to the Google Groups "Logisim" group.
> To post to this group, send email to log...@googlegroups.com.
> To unsubscribe from this group, send email to logisim+u...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/logisim?hl=en.
>
>

Ilia Lilov

unread,
Mar 3, 2011, 12:26:54 AM3/3/11
to log...@googlegroups.com
Your are right, Carl, there is no need to release new fragile with
those bugs fixed: those menu items don't have part in screen shots.
Are you going to cover those two pop-up menu items in the
documentation? I think it worth since it's quite unexpected: actually,
there is almost no more components with unique pop-up menu.

Very tiny GUI bug: after change from Edit Tool or Select Tool to Poke
Tool (both using Toolbar or using Explorer Pane) Edit Tool or Select
Tool keep blue background of icon (although black frame around icon
switches to Poke Tool correctly).

Ilia.

Ilia Lilov

unread,
Mar 3, 2011, 12:35:26 AM3/3/11
to log...@googlegroups.com
Another one GUI bug; to reproduce:

1) Right click on a component on the canvas, choose "Show attributes".
Blue oreol will appear.
2) Try to select any other component on the canvas. I will be able to
do that, but attribute pane will not update and oreol from previous
component will not disappear.
3) Change current tool to remove the oreol.

Ilia Lilov

unread,
Mar 3, 2011, 2:28:50 AM3/3/11
to log...@googlegroups.com
Bug:
Preferences window, Layout tab: value from "First radix when wire
poked" actually used for first radix and for second radix; "Second
radix when wire poked" value is ignored.

Ilia Lilov

unread,
Mar 6, 2011, 6:03:01 AM3/6/11
to log...@googlegroups.com
Hello, all.

There is an attachment with the file to demonstrate new bug. At the
main circuit there is a node at the combined end of the splitter. If
you will copy this splitter, it will have an "ability" to create this
unexpected node too. I don't know how to create such magic splitter
since this circuit came from one of my students.

Also, I have the suggestion. Resolution of screens as well as
resolution of projectors grows, so limit of 200% for circuit scale
isn't enough to have full-screen demonstration of arbitrary circuit on
some projectors. So I suggest to increase this limit.

Carl, since you added enable input pin for MUX and DEMUX, many
circuits containing them work incorrectly, because horizontal wire
coming to control input connects to enable input too - automatically
and silently. I don't suggest to remove enable input of course, but
may be it is possible to detect such connections (enable input
short-wired to control input) and to show notification messages with
name(s) of circuits where it occurred. For example these "Pong"
circuit which you sent to me at summer have this problem and it's
annoying to search such connections manually.

Ilia.

ZeroLengthWire-bug.circ

Ilia Lilov

unread,
Mar 6, 2011, 6:28:56 AM3/6/11
to log...@googlegroups.com
There are two files in the attachment to demonstrate quite huge bug in
Combinational Analysis module. There is truth table for 3-to-8 decoder
with enable input. The module will generate a circuit with mistake:
"e" and "c0" inputs are short-wired.
enabled-decoder-3-to-8.txt
enabled-decoder-3-to-8.circ

C Burch

unread,
Mar 6, 2011, 10:14:08 PM3/6/11
to log...@googlegroups.com
Thanks, Ilia, for the report. The "zero-length" wire here actually
isn't zero length: The student managed to add a minimum-length
(10-pixel) wire that connects the combined end to the splitter's
"spine." Usually Logisim repairs this silently when you create a wire
slightly overshooting the combined end, so it doesn't happen very
often. The student probably did something like this: First, draw a
wire across the combined end into the split end, and then shorten the
wire dragging from the split end to the spine.

In any case, this isn't technically a bug. It could some day prove to
be enough of a source of confusion to merit some special feature to
detect it. But this is the first I've heard of confusion over the last
five years, so I'm not counting it as very urgent - particularly it's
a lot less likely to occur with the new splitter appearance.

-Carl

Reply all
Reply to author
Forward
0 new messages