[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Savannah-hackers-public] Project tops showing old files

From: Bob Proulx
Subject: Re: [Savannah-hackers-public] Project tops showing old files
Date: Sat, 22 Jun 2019 21:20:17 -0600
User-agent: Mutt/1.10.1 (2018-07-13)

Hello Dale,

Dale Williamson wrote:
> I believe I have narrowed down what is probably the machine that made
> the most recent commit.  In our commit log it was on Thu Jun 13 15:04:52
> PDT 2019 for file viewvc/tops/tops/sys/math.v.

That's all good information but I am hoping that was from before your
report and before I updated the "tops" project and sync'd in those
commits.  When I wrote my previous message I had already determined
that the tops project had newer commits in the nfs1 version of the
repository.  I had therefore sync'd the CVS source repository forward
from there to the live production copy.

The live copy since my previous message should be completely correct
for the tops project.

> On your system I just viewed in a browser that file, and it shows:
>    Log of /tops/tops/sys/math.v
>    Revision 1.6.2 Thu Jun 13 22:07:23 2019 UTC

CVS (and RCS) revisions are always pairs of numbers.  So 1.6.2 is
perhaps 1.62?

> This time is consistent with the log time (PST) above.  In the browser I
> was able to view "Diff to previous 1.6.1" and it matches the changes
> made.
  Revision 1.62 - (view) (download) (annotate) - [select for diffs] 
  Thu Jun 13 22:07:23 2019 UTC (9 days, 5 hours ago) by dwil

> But on your system viewvc/tops/tops/sys/?sortby=date shows for file
> math.v: Rev 1.54 Age 2 months.  In fact, many files show age 2 months.
> That is at odds with what I just viewed in a browser as Rev 1.62 Thu Jun
> 13 22:07:23 2019 UTC.

I am hoping that we were just working in parallel and so our emails
passed each other in flight.

> While this discrepancy on your system exists, we won't make any updates
> or commits from our machines since their CVS/Entries will probably not
> square with yours and could cause a huge mess.

Thank you!  It would definitely cause a problem similar to the
two-brain problem if commits were made to both repositories
separately.  Because then the same revisions of a file could have two
different sets of changes.  However for your project "tops" I think
everything should be okay.

In particular I compared the CVSROOT/history file which contains every
action that has happened to the two repositories.  I could see that
the commits you had noticed that had gone missing were in the nfs1
system's version but not the vcs1 systems version.  vcs1 is the
current live production copy.  Therefore I sync'd the nfs1 version
over to vcs1 so all should be good for the "tops" project now.

> Wishing you best of luck and awaiting word of your success.

I have been analyzing the repositories.  It looks like I was quite
lucky in that there were no occurrences of commits happening to both
versions after the failed sync.  It looks like the web pages were
sync'd correctly.  It looks like there were 18 source projects with
commits to the nfs1 version that were failed to copy back.  Your
project "tops" was one of those 18.  I have sync'd tops.  So you
should be good to go.  That leaves only 17 projects for me to
carefully review manually and to sync forward.  And if I can do so
quickly I may avoid the messiest of the problems.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]