cvs2git - branches without sprout and related git orphan commits

Hi,

Our cvs2git conversion is resulting in a git repo with multiple orphan 
commits (commits without any parents at all and are not the initial 
commit).  I can't find any issue tracker issues or mailing list threads 
with exactly this same problem.  We're using cvs2git from cvs2svn 2.4.0 
(22 September 2012).

The resulting repo has an initial commit from 2001, but also ends up 
with 2 orphan commits from 2006 and 2014:


[builder@stlbldd-new smsmis]$ git log --remotes=origin/branch-* 
--reverse --date=local --pretty=format:"%h parents(%p) %ad %an %s" | 
grep "parents()"
c837450 parents() Mon Jul 23 18:50:02 2001 jdoyle Added starter project 
file and package hierarchy for diagram files.
56154d6 parents() Fri Jan 27 17:54:32 2006 jdoyle SIR #: (none) 
Description: Added new build config files for new asmp2-check test area 
used for ASM Phase 2 Checkout.
8a97b86 parents() Thu Oct 16 12:50:50 2014 jdoyle SIR #: (none) 
Description: New Data Mapping document for Divisional Accounting Release 
3. In progress. Completed


These orphan commits cause the history to look very odd for all the 
files that existed during that time period, e.g. when doing "git log 
--graph".

When I look at git-dump.dat, I can locate those commit entries and see 
that they are the first commit for the respective branch, but do not 
have a "from :NNNNN" line, e.g.:


commit refs/heads/branch-div-account-3-dev
mark :1000062509
committer jdoyle <> 1413478250 +0000
data 246
SIR #: (none)
Description: New Data Mapping document for Divisional Accounting Release 3.
In progress. Completed basic algorithm for DaOnlyBidsBean. Need to add
algorithm steps for populating default subaccounts and for any 
validation desired.

M 100644 :71793 sms/docs/population/Bids Data Mapping.docx


There are later manufactured commit entries for the branch but their 
message only says "Cherrypick" and not "Sprout", and they don't have a 
"from :NNNNN" line either:


commit refs/heads/branch-div-account-3-dev
mark :1000062536
committer cvs2svn <[email protected]> 1413894819 +0000
data 750981
This commit was manufactured by cvs2svn to create branch 'branch-div-
account-3-dev'.

Cherrypick from master 2014-10-21 12:33:38 UTC npatel 'SIR #: (none)':
...
Cherrypick from master 2014-10-10 14:38:23 UTC builder 'Baseline output 
export generated with SMS/MIS trunk build 2014-10-10 11:48:25 GMT':
...
    data-exports/baseline-output/sms/SERVICE_SETTLEMENT_T.sql
M 100644 :79404 
sms/resources/config/producttest/testapps/pnodevirtualbidproxyconfig-application.xml
...


All the other branches seem to start with a manufactured commit entry 
with a "from :NNNN" line and a commit message that says "Sprout from", e.g.:


commit refs/heads/branch-37_0_X
mark :1000073083
committer cvs2svn <[email protected]> 1488218199 +0000
data 19007
This commit was manufactured by cvs2svn to create branch 'branch-37_0_X'.

Sprout from master 2017-02-27 17:56:36 UTC npatel 'SIR #: (none)'
Delete:
CVSROOT/checkoutlist
CVSROOT/commitinfo
...
data-exports/baseline-output/sms/VAR_STANDARD_STLMNT_T.sql
from :1000073080
M 100644 :88564 
sms/src/htdocs/metering/MeteringUI/app/view/fcmdemandassets/MainModel.js
...
M 100644 :147297 sms/test/database/model/CONTRACT_T.sql
M 100644 :57575 mis/src/database/views/MIS_DAILYPOS_SCHED_2_VAR_V.sql
D data-exports/baseline-output/sms/ENERGY_RT_STLMNT_T.sql
D data-exports/baseline-output/sms/NCPC_RT_ARD_CRDT_SP_DTL_T.sql
...
D data-exports/baseline-output/sms/RSV_CUST_LZ_T.sql
D CVSROOT/checkoutlist


It kind of seems like for these 2 branches, cvs2git is waiting to create 
a manufactured commit for the branch until too late, after a regular 
user commit has already been created.

I think this is probably related to our cherrypicking the branch 
contents in CVS, and I could alter the CVS repository files if it would 
help, but I can't pinpoint what it is about these 2 branches that 
doesn't work, when there are 100s of other branches that seem to have 
been OK.

What is the best way to troubleshoot what's going on?

Thanks,
Jim

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=1670&dsMessageId=3263939

To unsubscribe from this discussion, e-mail: [[email protected]].