[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference smurf::buildhelp

Title:USG buildhelp questions/answers
Moderator:SMURF::FILTER
Created:Mon Apr 26 1993
Last Modified:Mon Jan 20 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2763
Total number of notes:5802

51.0. "revision comments getting rearranged ..." by SMURF::FILTER (Automatic Posting Software - mail to flume::puck) Tue May 18 1993 16:39

Date Of Receipt: 	14-MAY-1993 13:27:53.78
From: 	LOCORE::"[email protected]"
To: 	[email protected]
CC: 	decwet::mansour, decwet::thomas, [email protected], [email protected]
Subj: 	revision comments getting rearranged ...

  The problem that I tried to solve in the patch of bsubmit that you got,
  is the following:

  in a lot of cases, detected merge conflicts were caused by headers 
  conflicts. And this can apply, to projects divided into,
  shared sandboxes, stable backing tree, and submit trees.

  When users, are submitting a fix to a submit tree, and they 
  want to submit it, to the shared sandbox that is part of the project too,
  they usually got into unnecessary merge conflicts, due to comments
  and revision  number existing on one branch (ex. submit tree),
  but not in the other(ex. shared sandbox).

  The approach used , was, to extract history revision,  from 
  user private branch revision, submit public branch revision and 
  the ancestor of both private and public versions of the same file,
  and group them by submit branch, and other public branch,
  like shared sandbox branch,  and sort each group by revision number, 
  in decreasing order.

====> This is why you noticed the reshuffling of comments into the log
      history.

  Therefore, we decided to add this feature to eliminate the risk
  of dealing with merge conflicts caused by history log.
  There would be no differences in history log detected on newly
  created file, but they would be some detected on already existing 
  files for the first time only; and this due to the sorting and 
  grouping of history log.

 ( DETAILS:

  The significant components of headers are stored in order into
  two linked lists, one for submit branch, and the other is for different
  public branch. And then both linked lists are read into a
  file, resulting in two sorted groups of headers history by revision numbers.

  So, users can see always the latest changes that happen to the file,
  on top of the history.
  ).


  N.B.  I'd like to add that merges in bsubmit are only being done 
        at the level of code or body of the file now. 
        So, no merges will be done at the level of history revisions. 
        The merged code will be appended to the sorted history.
 
  So, if you feel that you really don't need this feature now,
  please let us know and we can back out the changes.

 
thank you,

/Rima



T.RTitleUserPersonal
Name
DateLines
51.1fwd: revision comments getting rearranged?SMURF::FILTERAutomatic Posting Software - mail to flume::puckTue May 18 1993 16:38532
Date Of Receipt: 	13-MAY-1993 18:33:09.00
From: 	MINSRV::"[email protected]" "Joshua M. Friedman ULTRIX SDE"
To: 	[email protected]
CC: 	[email protected], [email protected]
Subj: 	fwd: revision comments getting rearranged?

Rima (et al),

Jim Campbell sent the attached question about rearrangement of headers
now in ode.  Yesterday John Flanagan found this sort of case, but which
included branch numbers from a different rcs pool (our old mips pool)
in with branches from the current os pool.  This added additional
confusion.  In the attached diffs, the new header seems to have older
stuff (lower branch #s) closer to the top. ???

Attached below Jim's mail is the ptt history of a bug fix which you
just patched and which we have incorporated in our environment, which
is the cause (I assume) of this rearrangement.  I can see the value
in cleaning up headers, and little by little files which have, in
some sense, been jumbled up, will get "fixed" and more orderly.

Jim's feeling on this was that files which already have history should
be left alone, since diffs are confusing and unexpected.  He was
wondering if there was some way we could or should disable this feature
until it gets more discussion or concensus, since it presents major new
diffs to people.

Can you give us more information on exactly how things work now and
what the goal(s) were?  Thanks.  Your explanation in the attached ptt
report isn't detailed enough to really understand exactly what to
expect, although the scenarios presented are very explicit:

>   a possible solution can be ordering and sorting all revisions log history
>   revision in a ascendent order, with their correspondent comment.
> 
>   So in this way, we can keep track of all modifications done to the file,
>   and eliminate the occurences of header conflict.

>  the problem described above is being solved in DECode version 2.1.
>  the merge routine called by bsubmit command, is being modified to eliminate
>  any header conflicts that used to cause merge conflicts.

Maybe if you could provide an explanation (like patch release notes,
available as part of the ode_2.0_bug_fix sup collection) which we could
provide to our users ('usg') of what to expect, and how to appropriately
do merges with old and new header content, this would help us all.

Thanks very much....			--josh


--------

Return-Path: [email protected]
Received: by locore.zk3.dec.com (5.65/DEC-USSG-ZK3-ULTRIX-09/27/91);
	id AA01648; Thu, 13 May 1993 17:06:05 -0400
Message-Id: <[email protected]>
To: [email protected]
Cc: [email protected], [email protected]
Subject: revision comments getting rearranged?
Date: Thu, 13 May 93 17:06:04 -0400
From: "Jim Campbell dtn: 381-0792" <[email protected]>
X-Mts: smtp


Attached is an ode 'problem' I don't understand.

I did a submit of kn7aa.c. I deleted a printf, added a revision
comment and submited it. A two line change.

I did a diff just now of the file I bci'ed (I keep copies of every
submit around to ensure I can recreate them if things go wrong)
and what is in the nightly and there is a significant difference
in the revision comments section. 

The code portion is as I expected. (i.e. correct)

Attached is the diff. How did these comments get shuffled around?
It's quite disconcerting to do a diff of a two line change and get 200
lines that differ. This also happened to sched_prim.c.

Is there something I did to make this happen? It's bad when this happens,
since it will create a hugh amount of chaff when merges occur.

Regards, Jim...

7,52d6
<  * Revision 1.1.7.8  1993/05/11  21:52:17  Jim_Campbell
<  * 	Remove presto not present message during config.
<  * 	[1993/05/11  21:48:31  Jim_Campbell]
<  *
<  * Revision 1.2.4.9  92/04/02  13:47:07  Peter_Mott
<  * 	Define missing symbol PRM_DBG.
<  * 	Include missing routine ka_ruby_check_todr_setup.
<  * 	[92/04/02  13:46:23  Peter_Mott]
<  * 
<  * Revision 1.2.4.8  92/04/01  12:38:26  Peter_Mott
<  * 	Put in new readtodr and writetodr to use standard routines between
<  * 	cobra, ruby, et. al.
<  * 	[92/04/01  12:37:00  Peter_Mott]
<  * 
<  * Revision 1.2.4.7  92/03/17  14:16:37  Jim_Campbell
<  * 	First cut at OSFPAL (builds successfully). Fixed cfe warnings.
<  * 	[92/03/17  14:10:21  Jim_Campbell]
<  * 
<  * Revision 1.2.4.6  92/03/09  23:37:13  Jim_Campbell
<  * 
<  * 	Presto tweeks
<  * 	[92/03/09  22:12:05  Jim_Campbell]
<  * 
<  * Revision 1.2.4.5  92/02/10  13:40:40  Jim_Campbell
<  * 
<  * 	added latent presto support, fixed broken writetodr routine,
<  * 	fixed up indentation
<  * 	[92/02/10  13:21:10  Jim_Campbell]
<  * 
<  * Revision 1.2.4.4  92/02/07  18:08:57  Jay_Estabrook
<  * 	Alpha DSA merge
<  * 	[92/02/07  18:04:19  Jay_Estabrook]
<  * 
<  * Revision 1.2.4.3  92/02/05  17:54:51  Peter_Mott
<  * 	removed superfluous copyright statement.
<  * 	removed erroneous local declaration of yrtime in ka_ruby_writetodr.
<  * 	corrected indentation - tab stops set to 8.
<  * 	[92/02/05  17:34:21  Peter_Mott]
<  * 
<  * Revision 1.2.4.2  92/02/05  14:34:33  Peter_Smith
<  * 	Change names of time conversion routines.
<  * 	[92/02/05  14:32:59  Peter_Smith]
<  * 
<  * Revision 1.2  92/01/15  01:03:08  devbld_gsf
<  * 	Baselevel alpha_bl002
<  * 
60c14
<  * 
---
>  *
83,89d36
<  * Revision 1.1.3.2  92/08/21  05:18:50  Peter_Mott
<  * 	Add machine check parsing.
<  * 
<  * 	Add clock month bias to match that used by OSF.
<  * 
<  * 	Cleanup debug.
<  * 
184a132,262
>  * 
>  * Revision 1.1.2.9  92/07/09  09:53:57  Randall_Brown
>  * 	Cleanup of #ifdef OSFPAL
>  * 	[92/07/08  08:40:29  Randall_Brown]
>  * 
>  * Revision 1.1.2.8  92/06/23  16:41:18  Jim_Campbell
>  * 	Missed BL7 mirror: Initialize *system_intr_cnts_type_transl
>  * 	[92/06/23  16:07:55  Jim_Campbell]
>  * 
>  * Revision 1.1.2.7  92/06/03  20:33:43  Joseph_Amato
>  * 	badaddr interface changes and 1st pass of fbus
>  * 	[92/06/03  10:33:55  Joseph_Amato]
>  * 
>  * Revision 1.1.2.6  92/05/27  19:21:11  Jim_Campbell
>  * 	Just changed the boot message to VAX7000
>  * 	[92/05/27  19:05:37  Jim_Campbell]
>  * 
>  * Revision 1.1.2.5  92/04/24  15:36:51  Anton_Verhulst
>  * 	Add initialization of global pointer system_bus.
>  * 	System_bus is global pointer used by loadable code when walking
>  * 	the hardware topology tree.  It is extern'd via devdriver.h.
>  * 	[92/04/24  15:34:44  Anton_Verhulst]
>  * 
>  * Revision 1.1.2.4  92/04/20  15:36:06  Jim_Campbell
>  * 	Silver BL6 merge: modified presto interface
>  * 	[92/04/20  15:35:21  Jim_Campbell]
>  * 
>  * Revision 1.1.2.3  92/04/10  13:22:21  Stuart_Hollander
>  * 	errlog.h is now in dec/binlog
>  * 	[92/04/09  14:46:30  Stuart_Hollander]
>  * 
>  * Revision 1.1.2.2  92/04/03  12:37:43  Stuart_Hollander
>  * 	scb.h is now in arch/alpha
>  * 	ka_ruby.h now in hal
>  * 	Moved file from arch/alpha to hal
>  * 	[92/04/02  19:08:49  Stuart_Hollander]
>  * 
>  * 	Initial version.  File copied from TIN-based Alpha pool.
>  * 	[91/12/06  16:04:51  David_Snow]
>  * 
>  * Revision 1.1.2.8  92/01/14  13:19:54  Peter_Mott
>  * 	fix up todrread and write routines.
>  * 	[92/01/14  12:47:10  Peter_Mott]
>  * 
>  * Revision 1.1.2.7  92/01/02  11:38:44  Al_Delorey
>  * 	Merged my changes with prior version
>  * 	[92/01/02  11:36:40  Al_Delorey]
>  * 
>  * 	Fixed a comment (to test bmerge)
>  * 	[92/01/02  11:32:35  Al_Delorey]
>  * 
>  * 	No real source change, just a test of ODE-II bmerge
>  * 	[92/01/02  11:15:30  Al_Delorey]
>  * 
>  * Revision 1.1.2.6  92/01/02  11:10:12  Jim_Campbell
>  * 	Checked out to test bmerge functionality with Al...
>  * 	[92/01/02  11:09:27  Jim_Campbell]
>  * 
>  * Revision 1.1.2.5  91/12/30  18:49:49  Jim_Campbell
>  * 	Commented out debug messages, cleaned out some unused (adu) code
>  * 	[91/12/30  18:39:47  Jim_Campbell]
>  * 
>  * Revision 1.1.2.4  91/12/11  18:26:02  Al_Delorey
>  * 	scb.h moved to hal
>  * 	[91/12/11  18:24:17  Al_Delorey]
>  * 
>  * Revision 1.1.2.3  91/12/07  18:16:17  Al_Delorey
>  * 	fixed include paths for pool reorg
>  * 	[91/12/07  17:25:51  Al_Delorey]
>  * 
>  * Revision 1.1.2.2  91/12/06  17:54:11  David_Snow
>  * 	"Initial version.  File copied from TIN-based Alpha pool."
>  * 
>  * Revision 1.1.3.2  92/08/21  05:18:50  Peter_Mott
>  * 	Add machine check parsing.
>  * 
>  * 	Add clock month bias to match that used by OSF.
>  * 
>  * 	Cleanup debug.
>  * 
>  * Revision 1.2.4.9  92/04/02  13:47:07  Peter_Mott
>  * 	Define missing symbol PRM_DBG.
>  * 	Include missing routine ka_ruby_check_todr_setup.
>  * 	[92/04/02  13:46:23  Peter_Mott]
>  * 
>  * Revision 1.2.4.8  92/04/01  12:38:26  Peter_Mott
>  * 	Put in new readtodr and writetodr to use standard routines between
>  * 	cobra, ruby, et. al.
>  * 	[92/04/01  12:37:00  Peter_Mott]
>  * 
>  * Revision 1.2.4.7  92/03/17  14:16:37  Jim_Campbell
>  * 	First cut at OSFPAL (builds successfully). Fixed cfe warnings.
>  * 	[92/03/17  14:10:21  Jim_Campbell]
>  * 
>  * Revision 1.2.4.6  92/03/09  23:37:13  Jim_Campbell
>  * 
>  * 	Presto tweeks
>  * 	[92/03/09  22:12:05  Jim_Campbell]
>  * 
>  * Revision 1.2.4.5  92/02/10  13:40:40  Jim_Campbell
>  * 
>  * 	added latent presto support, fixed broken writetodr routine,
>  * 	fixed up indentation
>  * 	[92/02/10  13:21:10  Jim_Campbell]
>  * 
>  * Revision 1.2.4.4  92/02/07  18:08:57  Jay_Estabrook
>  * 	Alpha DSA merge
>  * 	[92/02/07  18:04:19  Jay_Estabrook]
>  * 
>  * Revision 1.2.4.3  92/02/05  17:54:51  Peter_Mott
>  * 	removed superfluous copyright statement.
>  * 	removed erroneous local declaration of yrtime in ka_ruby_writetodr.
>  * 	corrected indentation - tab stops set to 8.
>  * 	[92/02/05  17:34:21  Peter_Mott]
>  * 
>  * Revision 1.2.4.2  92/02/05  14:34:33  Peter_Smith
>  * 	Change names of time conversion routines.
>  * 	[92/02/05  14:32:59  Peter_Smith]
>  * 
>  * Revision 1.2.4.3  92/02/05  17:54:51  Peter_Mott
>  * 	removed superfluous copyright statement.
>  * 	removed erroneous local declaration of yrtime in ka_ruby_writetodr.
>  * 	corrected indentation - tab stops set to 8.
>  * 	[92/02/05  17:34:21  Peter_Mott]
>  * 
>  * Revision 1.2.4.2  92/02/05  14:34:33  Peter_Smith
>  * 	Change names of time conversion routines.
>  * 	[92/02/05  14:32:59  Peter_Smith]
>  * 
>  * Revision 1.2  92/01/15  01:03:08  devbld_gsf
>  * 	Baselevel alpha_bl002

------- Forwarded Message

Return-Path: [email protected]
Received: by wasted.zk3.dec.com (5.65/DEC-USSG-ZK3-ULTRIX-09/27/91);
	id AA04611; Fri, 7 May 1993 13:51:15 -0400
Date: Fri, 7 May 1993 13:51:14 -0400
Message-Id: <[email protected]>
From: [email protected] (07-May-1993 1051)
To: [email protected]
Subject: Results from: history ODE-00122

          Action Problem                                               Last    
Number    Status Status Pri   Owner        PRODUCT         Flags Age  Updated  
======    ====== ====== ===   =====        =======         ===== ===  =======  
ODE-00122 CL     CL     3     MANSOUR      ODE                  143 6-MAY-1993 
DEC - Merge Conflicts Caused By Headers Conflicts


Submitted by: RIMA MANSOUR              of ODE                       in USA       
	(548-8709            )
	Mail address: FUNGUS::MANSOUR

Answer code: FN
Fixed in version 2.1       

Last accessed on: 6-MAY-1993 09:46:   Number of days open: 143  
(RC: 28.9  AK:113.8  SB:  0.0  EL:  0.0  CL:  0.0)
(UN:142.7  DE:  0.0  RE:  0.0  ER:  0.0  KN:  0.0  ST:  0.0  CP:  0.0  TR:  0.0)	
Problem first entered on  14-DEC-1992

PRODUCT: ODE 2.0     
OS HARDWARE: ULTRIX/RISC 4.2     
COMPONENT: BSUBMIT 2.0     
OTHER:          

Hardware: RISC              	System device:           
Memory:   

Abstract: Merge Conflicts Caused By Headers Conflicts

Customer company: DEC             
[This update was sent from MOLD::"[email protected]" on 14-DEC-1992 17:13
[ODE responsible for next action, Problem is undefined]

Mail this report to:         [email protected]

Note: ALL CAPS=Mandatory field
Problem Submission Form  Mon Dec 14 17:05:34 PST 1992
Required Information
====================
YOUR NAME:                   Rima Mansour
Your Email address:          [email protected]
YOUR PHONE NUMBER:           548-8709
Your fax number:             DTN 548-8890

Your company name:           DEC
Your mailstop:               ZSO
YOUR ORGANIZATION'S NAME:    DECwest
YOUR COST CENTER:            35W
Your street address:         Digital Equipment Corporation
                             DECwest Engineering
                             14475 N.E. 24th St.
                             Bellevue, WA 98007 USA
- --------------------------------------
CUSTOMER'S COMPANY NAME:     DEC
CUSTOMER COUNTRY:            USA

- --------------------------------------
PRODUCT:                     ode
Product version (optional):  V2.0
OS HARDWARE:                 ULTRIX/RISC
OS Hardware version:         0
Workstation soft version (optional): V4.2
COMPONENT:                   bsubmit
Component version (optional): ode2.2b000:V-1.157:Prod-ver--V2.0
Other (optional):
OTHER Version (optional):
SYSTEM WHERE PROBLEM OCCURRED: rima.zso.dec.com

CPU type:                    RISC
Display type:                ? display of type=
Memory size:
System disk:
Other hardware configuration information:

Other software configuration information:
ULTRIX V4.2 (Rev. 96) System #1: Wed Nov 13 16:05:24 PST 1991
UWS V4.2 (Rev. 272)

- --------------------------------------
PROBLEM ABSTRACT (80 characters or less):
Merge Conflicts Caused By Headers Conflicts

PROBLEM PRIORITY (1-5, 1 being most critical): 3
CAN THE CUSTOMER REPRODUCE THE PROBLEM [Y/N]:  Y
CAN YOU REPRODUCE THE PROBLEM [Y/N]:           Y


PROBLEM STATEMENT:
=================

scenario 1:
- -------------

PRE:
- ----
 REV(submit_tree)(<filename>): defined
 REV(sharedsb)(<filename>): defined


    -------------              |       |
   | submit_tree |             |       |-----------------
    -------------              |       | 1.2.3:              private branch
         /\                    |       |-----------------
         ||                    |   T   |
         ||                    |   R   |-----------------
        -----                  |   U   | 1.2.2:              submit tree branch
       | SB1 |                 |   N   |-----------------
        -----                  |   K   |
                               |       |-----------------
 1) bco <filename>             |  1.2  | 1.2.1:              shared sandbox bran
ch
 2) bci <filename>             |       |-----------------
 3) bsubmit -noauto_out <filename>
 4) resb <shared_sandbox>
 5) bsubmit <filename>

  Initially, file(A) has been submitted to both the submit tree and the
  shared sandbox.

  After step 1), log history of file(A) contains revision:

   1.2             (initial rev),
   1.2.2.2         (first submission to the submit tree),
   1.2.3.1         (private branch rev),

  After step 3), log history of file(A) contains revisions:

   1.2             (initial rev),
   1.2.2.2         (first submission to the submit tree),
   1.2.3.2         (private branch rev after bci ),

  Since we choose -noauto_out, nothing changes after step 4) .

  At step 5), merge conflict are detected due to the conflict of log history
  revisions and comments, between my private branch and the shared sandbox
  branch :

  shared sandbox branch:
  ----------------------
   1.2       (initial rev)
   1.2.1.2   (first submit rev to shared sandbox)       <------------------
                                                                          |
  my private branch:                                                      |
  ------------------                                                  Conflict
                                                                in file revision
   1.2             (initial rev),                                         |
   1.2.2.2         (first submission to the submit tree),                 |
   1.2.3.2         (private branch rev after bci ),    <------------------|


scenario 2:
- -----------

    -------------
   | submit_tree |
    -------------
         /\
         ||
         ||
        -----
       | SB1 |
        -----

  1) bco <filename>
  2) bci <filename>
  3) bsubmit -noauto_out <filename>
  4) resb sharedsb
  5) bmerge -r"$default_set;<>" <filename>
  6) bci ...
  7) bsubmit <filename>


  In scenario 2, the same merge conflicts are detected at step 5), caused by the
  same header conflict discussed above.

  I would like to note that the same will happen, if we reverse the
  order of submitting to submit tree and shared sandbox.

/Rima


Instructions to reproduce the problem:
======================================



- --------------------------------------



	<<<<<<<<<<<<<<<<<<<<<<<PTT UNIT SEPARATION>>>>>>>>>>>>>>>>>>>>>>              

[This update was sent from PETERSON on 14-DEC-1992 17:22]
[ODE responsible for next action, Problem is undefined]
[This update was NOT sent via mail or to any gateway/bridge]

Problem has been reassigned to RIMA MANSOUR

	<<<<<<<<<<<<<<<<<<<<<<<PTT UNIT SEPARATION>>>>>>>>>>>>>>>>>>>>>>              

[This update was sent from RIMA::mansour on 15-DEC-1992 15:14]
[ODE responsible for next action, Problem is undefined]


 Hi,

  a possible solution can be ordering and sorting all revisions log history
  revision in a ascendent order, with their correspondent comment.

  So in this way, we can keep track of all modifications done to the file,
  and eliminate the occurences of header conflict.

/Rima

	<<<<<<<<<<<<<<<<<<<<<<<PTT UNIT SEPARATION>>>>>>>>>>>>>>>>>>>>>>              

[This update was sent from MANSOUR on  6-MAY-1993 09:44]
[ODE responsible for next action, Problem is undefined]







	<<<<<<<<<<<<<<<<<<<<<<<PTT UNIT SEPARATION>>>>>>>>>>>>>>>>>>>>>>              

[This update was sent from MANSOUR on  6-MAY-1993 09:48]
[The problem is closed]


   the problem described above is being solved in DECode version 2.1.
 the merge routine called by bsubmit command, is being modified to eliminate
 any header conflicts that used to cause merge conflicts.

thanks,

/Rima



------- End of Forwarded Message