roy@alfrat.uucp (Roy Phillips) (08/11/90)
With reference to your problem #5: GOOD segmentation violation
This problem is not an unknown phenomenon to me, below is a list of
three things worth checking:
1. Wrong paths in your environment - if older versions of Eiffel are
present, this can cause seg. viol.
2. Clusters/Classes of a previous version of Eiffel referenced in the
local '.eiffel' file (UNIVERSE: ...).
3. Certain crucial portions of an otherwise compiled system left
uncompiled - i've just eiffelclean'd and recompiled and everything
worked! (GOOD should handle un-compiled systems, but...).
The first thing to try is to create a trivial class containing, for example,
just an INTEGER feature, create the .eiffel with the smallest UNIVERSE, and
run good - this should always work.
Lastly, do you have 2.3 yet? the graphics library is far superior, in my view,
than the previous version - you might not have to mess around with the C
interface (urrgh!) at all, unless you're into serious 3D animation applications.
Tschuss!
Roy
-------------------------------------------------------------------------------
Roy Phillips | /~~~\/~~\ ,,, | roy@alfrat.de
A+F SystemEntwicklung | ( [ U U [||||||||||<<< | {world}!mcsun!unido!alfrat
D-4030 Ratingen | \___/\__\ ``` | "Don't Panic" - The Hitch-
West Germany Speak: +49-2102-850511 | -hikers Guide to the Galaxy
-------------------------------------------------------------------------------greg@trlamct.trl.oz.au (Greg Aumann) (08/13/90)
From article <104@alfrat.uucp>, by roy@alfrat.uucp (Roy Phillips): > With reference to your problem #5: GOOD segmentation violation > > This problem is not an unknown phenomenon to me, below is a list of > three things worth checking: > I have run accross this problem also. There is a problem when using GOOD with SUN's OpenLook server, at least with the 1.02beta version. It did however work with MIT's release 4 server. The same bug is demonstrated in the menu example program. See the execution trace below. I have reported this to ISE. I would like to see ISE publish a known bugs or problems list with patches or work arounds. This I am sure would save many people a lot of time. Also every now and again someone from ISE posts a patch. It would be very useful if this process was a little more formalised as it is in the source groups. By this I mean using context diffs so they can be safely applied automatically using patch(1) and including some sort of numbering or dating scheme eg patchlevels or the patch dates al la Cnews. --- tardis% demo Event type = 11 Internal error in function fill_input System execution failed. Below is the sequence of recorded exceptions: -------------------------------------------------------------------------------- Object Class Routine Nature of exception Effect -------------------------------------------------------------------------------- 1E465C GRAPH_SHELL execute Feature "handle": Applied to void reference. Fail -------------------------------------------------------------------------------- 1E465C GRAPH_SHELL execute Routine failure. Fail -------------------------------------------------------------------------------- 1E5E0C DEMO create Routine failure. Fail -------------------------------------------------------------------------------- tardis% --- To Christiane Greitmann - I have sent you some email but I am not all that confident that it will arrive. Please let me know if you don't get it. > The first thing to try is to create a trivial class containing, for example, > just an INTEGER feature, create the .eiffel with the smallest UNIVERSE, and > run good - this should always work. > > Lastly, do you have 2.3 yet? the graphics library is far superior, in my view, > than the previous version - you might not have to mess around with the C > interface (urrgh!) at all, unless you're into serious 3D animation applications. Does this mean that Eiffel 2.3 is released? If so when will those with support contracts get it? Perhaps someone from ISE could post the release notes. You say the graphics classes are usable does this mean that they now have a proper interface to X Windows and not this attempt at a generic windows interface complete with floating point co-ordinates? I consider that the lack of a good X windows user interface toolkit is, for me at least, one of the most serious deficiences of Eiffel and would be enough reason for me to use C++ on my next project despite the fact that I think Eiffel is a superior language. Greg Aumann ------------------------------------------------------------------------- Artificial Intelligence Systems Email: g.aumann@trl.oz.au Telecom Research Laboratories Voice: +61 3 541 6222 Melbourne, AUSTRALIA Fax: +61 3 543 8863
nosmo@eiffel.UUCP (Vince Kraemer) (08/14/90)
From article <2070@trlluna.trl.oz>, greg@trlamct.trl.oz.au (Greg Aumann): >From article <104@alfrat.uucp>, by roy@alfrat.uucp (Roy Phillips): >> >> Lastly, do you have 2.3 yet? the graphics library is far superior, in my view, >> than the previous version - you might not have to mess around with the C >> interface (urrgh!) at all, unless you're into serious 3D animation applications. > >Does this mean that Eiffel 2.3 is released? If so when will those with >support contracts get it? Perhaps someone from ISE could post the release >notes. You say the graphics classes are usable does this mean that they >now have a proper interface to X Windows and not this attempt at a generic >windows interface complete with floating point co-ordinates? > I will try to answer your questions as completely as I can: 1. Eiffel version 2.3 is currently in Beta release. It will be officially released to the general public some time in the future. All organizations that are eligible to receive 2.3 as an update will receive information about this fact. As usual, we will be posting a document outlining the changes, improvements, bug fixes and known bugs in the compiler and libraries about the time we release 2.3. Unfortunately, for Greg, the revisions to the graphics library have not included an encapsulation of the Xlib or Xt level of X, for this release. We are currently designing and implementing a more complete graphics library, which will have some of these low level elements. It will also have a much more complete set of UI objects. It will be released in ... the future. Vince Kraemer ISE Tech. Support (technical correspondence to: eiffel@eiffel.com)
nosmo@eiffel.UUCP (Vince Kraemer) (08/22/90)
From article <2070@trlluna.trl.oz>, greg@trlamct.trl.oz.au (Greg Aumann): > >I consider that the lack of a good X windows user interface toolkit is, >for me at least, one of the most serious deficiences of Eiffel and would be >enough reason for me to use C++ on my next project despite the fact that I >think Eiffel is a superior language. There are two independently developed UI toolkits available: The first is a Motif widget set encapsulation. It requires that you have the Motif widget set as part of your development environment. The toolkit comes as source, with some examples. Since it is an encapsulation, extendibility is limited. This toolkit will be part of the contributed software directory in the 2.3 release. The second toolkit implements an OpenLook'ing interface. It is written from scratch (the Xlib) level. It appears to be extensible, since so much of it is written in Eiffel. This toolkit has been written up in a proceedings on the European Unix User's Group, in 1989 (or 90). (I can not find my copy of it to give you all the details. I will keep looking). According to Andrea Querci, this toolkit has been available since May 1990. The Motif-based toolkit has been developed by: Intelligent Visual Computing P.O. Box 418 Morrisville, NC 27560 tel: 919-481-1353 fax: 919-467-3683 The OpenLook'ing toolkit is produced by: Unirel Centro Commerciale Osmannoro Via Tevere, 14 50019 Sesto Florentino ITALY tel: ++39-55-373043 fax: ++39-55-318525 (As an aside, to independent software developers: If you watch this NG at all, you realize there is a great demand for class libraries in a number of areas. Many of these could be quite lucrative ventures which ISE doesn't have the time or expertise to write quality class libraries for. By taking advantage of the C package options available, developers could conceivably write a package of C callable routines - if worries of having a limited market are slowing you down, maybe you should think again.) Vince Kraemer ISE Tech. Support (technical correspondence to: eiffel@eiffel.com)