Conference smurf::buildhelp

2708.0. "More on Krishna s bsubmit troubles..." by AOSG::FILTER (Automatic Posting Software - mail to flume::puck) Wed Dec 04 1996 10:03

Date Of Receipt: 	 4-DEC-1996 09:44:09.69
From: 	HUNCH::gillett "Christopher Gillett USG  04-Dec-1996 0940"
To: 	[email protected]
CC: 	
Subj: 	More on Krishna's bsubmit troubles...

------- Forwarded Message

Return-Path: gillett 
Delivery-Date: Wed, 04 Dec 96 09:35:57 -0500
Return-Path: gillett
Received: from ohunch.zk3.dec.com by quarry.zk3.dec.com; (5.65v3.2/1.1.8.2/16Jan95-0946AM)
	id AA31279; Wed, 4 Dec 1996 09:35:56 -0500
Received: from sledgeomatic.zk3.dec.com by hunch.zk3.dec.com; (5.65v3.2/1.1.8.2/11Mar96-0342PM)
	id AA10150; Wed, 4 Dec 1996 09:35:55 -0500
Received: from localhost by sledgeomatic.zk3.dec.com; (5.65v3.2/1.1.8.2/28Jul95-0202PM)
	id AA19348; Wed, 4 Dec 1996 09:35:54 -0500
Message-Id: <[email protected]>
To: [email protected]
Cc: [email protected]
Subject: Re: Looks like major problem with my _bsubmit_.. 
In-Reply-To: Your message of "Wed, 04 Dec 96 19:07:10 +0530."
             <[email protected]> 
Date: Wed, 04 Dec 96 09:35:54 -0500
From: gillett
X-Mts: smtp


Hi Krishna...

I'll take a closer look at the log you sent and see if I
can understand what is wrong.  I will also check with the
ODE support staff here and see if they understand as well.

I have some suggestions for making things easier for you.
If you are working on multiple bugs involving the same
file such that you will be making more than one set of 
changes to fix the bug, consider using multiple sandboxes.

For example, if I have 3 bugs to fix that all involve the
same file (say for example foo.c), I'll create 3 sandboxes
and check out foo.c in each one.  I'll fix each bug 
independently, and bci/srequest each bug separately.  Then
when I bsubmit, ODE will be helpful and do the merges
correctly.  This avoids troubles like you've been having
with diff code from other fixes showing up in current
srequest.

Next, you MUST bci before you srequest.  ALWAYS do this.

bcs command to unlock a file should only be used (in my
opinion) if you are going to throw away changes

I don't think you should be editing/creating .BCSconfig file.
That should be created automatically by ODE.  It's absence
means that it either got deleted, or that something is
wrong with ODE.

I looked at the steelos submit tree.  There is evidence that
your submit did something, but I'm not entirely certain what.

For example, the only differences in symtab.c are the log
file messages.  Same story with errors.list.

expr_sem.c has been changed.  I diff'd it against an older 
version and the diff follows:

diff expr_sem.c /usr/staff/gillett/scatbox/steel/src/usr/ccs/bin/cfe/expr_sem.c
7,10c7
<  * Revision 4.3.30.3  1996/12/04  12:04:07  Krishna_Babu_Bangera
<  * *** empty log message ***
<  *
<  * Revision 4.3.30.2  1996/05/10  19:22:31  Yakov_Yazlovitsky
- ---
>  * Revision 4.3.28.2  1996/04/15  14:53:43  Yakov_Yazlovitsky
16,17c13
<  *    [1996/04/15  14:53:43  Yakov_Yazlovitsky]
<  * 
- ---
>  *
279c275
< #pragma ident "$Header: /usr/sde/osf1/rcs/os/src/usr/ccs/bin/cfe/expr_sem.c,v 4.3.30.3 1996/12/04 12:04:07 Krishna_Babu_Bangera Exp $"
- ---
> #pragma ident "$Header: /usr/sde/osf1/rcs/os/src/usr/ccs/bin/cfe/expr_sem.c,v 4.3.28.2 1996/04/15 14:53:43 Yakov_Yazlovitsky Exp Yakov_Yazlovitsky $"
981a978,990
>       expr = root;
>       root = expression(root, context, lrvalue);
>       CHECK_ERROR_MARK_VOID(root,root);
> 
>       if (!IS_CONST_EXPR(root))
>       {
>           error(SEM_BAD_CONST_EXPR, Error, TREE_LOCATION(expr));
>           TREE_CODE(root) = Error_mark;
>           break;
>       }
> 
>       break;
> 
987c996,1001
< 
- ---
> #if 0
>     if( TREE_CODE(root) == Div_expr || TREE_CODE(root) == Rem_expr )
>       if( IS_CONST_EXPR(opr_1) && TREE_TYPE(opr_1) == int_type )
>         if( INT_CONSTANT(opr_1) == 0 )
>           root = zero_constant;
> #endif
995a1010,1011
> 
> 

I don't suspect that ODE actually *deleted* your files.  It does remove
them after a succesfull bsubmit (because you said 'Yes' to the question
"Outdate files if succesfull submission?")

You need to look at the diffs from expr_sem.c and decide if you've damaged
ths file with the bsubmit.  It's not clear to me what you did to the
file in this case.  

I'll forward your email to the ODE folks and see what they have to say.
If you get stuck with ODE problems, you can send email to [email protected].
That's an alias of individuals in engineering who are ODE experts (they
work on ODE, release engineering issues, and other nasty stuff full time).
They've saved me more than once. :-)  Hopefully they'll respond to you
sometime today, so that after you read this mail from me they'll be 
something from them.

I am out of the office from 5-December to 12-December.  Did you get my
earlier email this week about the documents being shipped?   If not, I 
sent the Pro Forma Invoice and Chartered Engineer's Certificate to b'lore
via DHL courier to attention "Krishna Bangera" so you should have it
in hand by close of business this week.  We await the consent of the
Software Technology Park officials and the Indian government to commence
shipping the workstation to you.

Thanks,
Chris



------- End of Forwarded Message