2.1b upgrade – include
13 messages in this thread
Greg,
Is it necessary to run the new include program when upgrading to 2.1b?
There is no mention of it, but I noticed that there was a newer version. I
tried it, and it crashed my system.
— Lowell
No, no reason to run the new include program when upgrading to Comeau C++
2.1b.
Re something crashing something: please, always give detailed details!
OK, here are the details…
I ran the 2.1 installation program without problems. Then I un-LHARC'ed
the 2.1b file, and executed the specified script (which has an error, by
the way, it tries to copy alibc21c.lib instead of alibc21.lib). I then ran
the 2.1b include program (which I figured was probably left out of the
instructions since there was a new one in the 2.1b LZH file), which ran for
a while, listing the things it was converting. In the middle of running,
it stopped, and the system requester "Software Error – Task Held" popped
up, which of course led to the Guru.
After this, I realized that I forgot to run the 2.1 include program, and
thought that might be related to the crash. So I cleared everything off
and started over. This time, after installing 2.1, I ran the 2.1 include
program. It also crashed the system. I figured out that I had not made
all the CC*: assigns, so I did so and tried it again. This time the 2.1
include worked. (It doesn't seem very user friendly to me to crash the
machine just because some logical devices aren't assigned.) Then I did the
2.1b installation. I tried the 2.1b include program again, and it crashed
again in the same manner (and this time I had the logicals assigned). I
doesn't seem to matter what I do, the newer include program always crashes
the machine. Since you say I really don't need to run the newer include,
I'm all set up. But the crashing obviously means something is wrong.
p.s. Have any luck with that execute script you were making up for me?
— Lowell
I'm a bit confused. include (the 2.1 or 2.1b version) does not depend upon
any assigns. Do you mean you specified paths on the command line with
assigns in them?
My versions of include (2.1 and 2.1b) replies with the following if I enter
it with no arguments (hopefully CIS won't mess it up):
WARNING: include path's are missing
Usage: include drive:comeau_include_path drive:c_include_path
Using CCINCLUDE and INCLUDE instead…
Target Comeau C++ directory='(NULL!!!)'
C compiler include files location ='(NULL!!!)'
Aborting: CCINCLUDE or INCLUDE empty.
You tell me what INCLUDE and CCINCLUDE are…
David – via Whap!
The same source is used for MS-DOS and AmigaDOS. They are environment
variables (CCINCLUDE being the Comeau C++ include directory, and INCLUDE
being the C include directory). The key here is that the include path's
are missing and even an attempt at checking the environment to let you do
something with subsequent confirmation, it can't. This is not the
suggested approach to running include.
Forgive me, Greg, but there is no "suggested" approach to running
include in the documentation that I got (I just checked). The 2.1b
upgrade makes a reference to changes that have been made in the
include program with the assumption that you (the user) knew how to
use it before. The Amiga paper documentation makes reference to
Install_Comeau_CPP using the include program to update your C
compiler headers, but that's not set up for Aztec C or for taking
apart the 2.1b archive (which has the Aztec support). BTW, it talks
about CC21include rather than CCINCLUDE (adding to the confusion).
There is no real statement on what include is attempting to do and how
far it will attempt to go (although, its probably possible to make an
educated guess) in case it fails to perform because the user had some
special set up and needs to do the set up manually.
>forgive me…
Actually, forgive me: the AmigaDOS addendum appears to be missing
a paragraph in the printed version. The last paragraph talks about
"In this step", but it is only discussed and never shown exactly what "this"
step was. Thanks for pointing this out.
Um, in terms of your immediate needs/concerns on the include command,
do you have any additional questions?
Short term concerns for 'include'? Arguments it takes and environment
variables it falls back on (I think we've covered that to some degree).
What type of environment variables it uses (Commodore/Manx – what command
do you use to create them?). Why it might fail (for instance, an empty
subdirectory seemed a problem for me). How the include mechanisms of the
various compilers work together (C++, SAS/C, DICE, and Aztec C) so that I
can ensure I'm set up correctly (I'm not sure about ComeauAssigns or what
Install_Comeau_Cpp did). Am I making sense?
David – via Whap!
On one try I said:
include cc21include: include:
but I had not run the ComeauAssigns script to define cc21include: yet, and
it crashed the machine. After I executed the assignment script, it worked
fine (the 2.1 version). The 2.1b version crashes the machine in all cases.
Very odd. All we do is an open. You should be getting a "blocking"
requestor about that, and even if you cancel I believe we catch it.
Let's see…
Ok, I just ran it with non-assigned logicals and it gives me a
"Please insert volume include in any drive". Clicking retry just
reissues the requestor. Clicking cancel produces an error message
of: "C compiler include directory 'include:' not valid!". This is
under 2.0. Let me reboot under 1.3…. well, that does the same
thing. I'm not very familiar with any of the add on tools to ADOS.
Do you have requestors or something like that disabled??
Sorry for taking this long to respond. I haven't been able to log on
recently due to heavy overtime at work.
To answer your question about having requestors disabled – no, I don't. I
have a fairly plain 1.3.2 system. The only real extras I have are Dmouse
(a mouse accelerator and screen blanker) and a tool to let me access
certain things from a drop-down workbench menu.
One possibility just occurred to me. Does that stack size of 100,000
required for the compiler apply to the include program also?
Did you ever come up with that basic execute script you said you were
making for me a while back? I haven't been able to try the compiler itself
yet, since I don't have Arexx or 2.0 yet, and can't even find anyone who is
still selling Arexx anymore.
— Lowell
>Sorry for taking this long to respond.
Ditto!
>To answer your question about having requestors disabled – no, I don't.
Hmmm…
>One possibility just occurred to me. Does that stack size of 100,000 required
>for the compiler apply to the include program also?
I'm pretty sure we tested that and it did not effect anything. I could be
wrong though…. Well, I just tried it with a 4K stack under 2.0 and got the
requestor.
>Did you ever come up with that basic execute script you said you were making
>for me a while back? I haven't been able to try the compiler itself yet,
>since I don't have Arexx or 2.0 yet, and can't even find anyone who is still
>selling Arexx anymore.
Hawes is still selling it as far as I know. I believe he has an id
here. Re the script, I thought I'd e-mailed about it a few weeks ago.
In any event, I will binary attach some mail to you. Please tell me if you
don't get it after seeing this message.