Bugzilla Bug 195429

elogv crashes with the following info:

Traceback (most recent call last):
  File "/usr/bin/elogv", line 447, in <module>
    curses.wrapper(ElogViewer)
  File "/usr/lib64/python2.5/curses/wrapper.py", line 44, in wrapper
    return func(stdscr, *args, **kwds)
  File "/usr/bin/elogv", line 174, in __init__
    self.refresh_file_pad()
  File "/usr/bin/elogv", line 256, in refresh_file_pad
    self.file_pad.refresh(self.pposy,0,1,1,self.height/2-1,self.width-2)
_curses.error: prefresh() returned ERR


Reproducible: Always

Steps to Reproduce:
1. Start elogv. You have to have more entries than can fit in one screen. 
2. Go to last entry and delete one after one.
3. When cursor will come to the first one (from top), elogv crashes.

Created an attachment (id=133087) [edit]
emerge --info

Adding author

Finding this bug makes me consider moving to an alternative elog viewer. I
filed a bug upstream for a related issue on September 22 and still no response
from the devs:

http://sourceforge.net/tracker/index.php?func=detail&aid=1800105&group_id=176946&atid=879268

Sorry that this took so long, but I am not very keen of Python, so I needed to
jump a high fence until I dared touching it.  Anyway, I fixed it by the cost of
some more screen redraws, emerge 0.6.1-r1 in about an hour, test it and report
back.  Thanks for the patience and for the report.

Upstream is mostly dead.  That's really sorry.

(In reply to comment #4)
> Sorry that this took so long, but I am not very keen of Python, so I needed to
> jump a high fence until I dared touching it.  Anyway, I fixed it by the cost of
> some more screen redraws, emerge 0.6.1-r1 in about an hour, test it and report
> back.  Thanks for the patience and for the report.
> 

Hi, thanks for Your try. However, Your solution 

self.change_usel(-1)
self.change_usel(+1)

doesn't work best for removing first file (change_usel(-1) doesn't do anything
and +1 moves to next row).

Finally, I've passed all exams, so I have some time and I looked at the code
(I'm not Python guru either) and tried to fix this issue. I added a check if
self.pposy > self.usel.

Moreover, after deleting log file, new log wasn't shown, so I added
self.show_log().

I don't think my solution is perfect, so please review my patch.

BTW: What do You think about adding vi-like j/k movement keys? Now it's just
arrow up/down...

> Upstream is mostly dead.  That's really sorry.
> 

Yeah, sad, I think console based elog viewer is really needed.

Created an attachment (id=142622) [edit]
patch against 0.6.1-r1

If You want, I can create patch against 0.6.1.

(In reply to comment #5)
> doesn't work best for removing first file (change_usel(-1) doesn't do anything
> and +1 moves to next row).

 Right, haven't thought about that corner case. :)

> Moreover, after deleting log file, new log wasn't shown, so I added
> self.show_log().

 Yeah, made me think about the solution for another bug.

> I don't think my solution is perfect, so please review my patch.

 Better than mine.

> BTW: What do You think about adding vi-like j/k movement keys? Now it's just
> arrow up/down...

 Nothing, as I am an Emacs user (and even maintainer). :)  But I added it now
in -r2.

> > Upstream is mostly dead.  That's really sorry.
> Yeah, sad, I think console based elog viewer is really needed.

 It still works and I can still patch it.  But it is the same with elogviewer,
kelogviewer and elogv (I maintain all three), all are dead upstream.  That's
the reason why developers are reluctant to add user-maintained programs...most
are unreliable and in the end one has to fix everything oneself.

Luca, if you are still reading this, have a look at the patches I added for
0.6.1-r1.  Thanks.

(In reply to comment #6)
> If You want, I can create patch against 0.6.1.

 Not needed, I did it myself.

Created an attachment (id=142684) [edit]
elogv-0.6.1-refresh-screen.patch

Hi at all,

Excuse me for responding too late, but I'm very busy in this period. By the
way, for the refresh-screen bug I found another solution that doesn't need to
rewrite the entire screen, it seems to work on my tests. Please try if it works
also on your systems. The other two patches (vi_movement, segfault-remove) will
be merge d on svn and on the next version that will be released soon. Thanks
for your work. ;)

And excuse me again, Luca

> Excuse me for responding too late, but I'm very busy in this period. By the
> way, for the refresh-screen bug I found another solution that doesn't need to
> rewrite the entire screen, it seems to work on my tests. Please try if it works
> also on your systems. The other two patches (vi_movement, segfault-remove) will
> be merge d on svn and on the next version that will be released soon. Thanks
> for your work. ;)
> 

 Worked into the ebuild -r3.  Thanks.