Mailing-List: contact dev-help@subversion.apache.org; run by ezmlm
Precedence: bulk
Received-SPF: neutral (nike.apache.org: local policy)
References: <1412157162.51628.YahooMailNeo@web87703.mail.ir2.yahoo.com> 
Message-ID: <1412180034.95203.YahooMailNeo@web87704.mail.ir2.yahoo.com>
Date: Wed, 1 Oct 2014 17:13:54 +0100
From: Julian Foad <julianfoad@btopenworld.com>
Reply-To: Julian Foad <julianfoad@btopenworld.com>
Subject: Re: Official way to create an empty revision
To: "dev@subversion.apache.org" <dev@subversion.apache.org>
Cc: Daniel Shahaf <d.s@daniel.shahaf.name>
In-Reply-To: <1412157162.51628.YahooMailNeo@web87703.mail.ir2.yahoo.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

I should have said, too: I'm not planning on taking this any further, mysel=
f. Anybody that wants to... feel free. (Discuss on this list first, as usua=
l, of course.)=0A=0A- Julian=0A=0AJulian Foad wrote:=0A> Daniel Shahaf wrot=
e in the thread "No no-op changes":=0A>>=A0 Should we provide an "official"=
 way to create an empty revision? That=0A>>=A0 is, a revision whose changed=
-paths list is empty?=0A>> =0A>>=A0 Use-cases:=0A>> =0A>>=A0 1. Suppose las=
t backup is r100 and revisions r101:r105 were lost; then=0A>>=A0 after rest=
oring the backup, the admin would create 5 empty revisions.=0A>> =0A>>=A0 2=
. Force an empty revision for whatever reason, such as to make the=0A>>=A0 =
revnums sync to something:=0A>>=A0 2.1. See r3 of the regression test merge=
_tests.py#125 svnmucc_abuse_1().=0A>>=A0 2.2. W hen loading our repository =
to the ASF repository, if Joe had=0A>>=A0 created 26 empty revisions, then =
The Offset would have been 840100=0A>>=A0 rather than 840074, which would m=
ake our mental math easier.=0A> =0A> Hi Daniel. It seems a reasonable tool =
to have in the svn admin's tool kit. =0A> Perhaps not often, but people do =
sometimes want this. I found two web pages =0A> where people discussed this=
. One wrote a script that spits out the appropriate =0A> few lines of dump =
file text to represent an empty rev, N times [1]; the other is =0A> worse, =
committing N changes to a temporary repo, dumping it and filtering =0A> eve=
rything out [2].=0A> =0A> I'm assuming this proposal is restricted to the a=
dmin side. Your use cases =0A> 1. and 2.2 are both admin use cases. Your us=
e case 2.1 is a test which uses a =0A> client-side commit to make an uninte=
resting revision, in order to make the =0A> subsequent revision numbers mat=
ch (modulo 10) those in the original use case. =0A> While people no doubt d=
o this sort of thing sometimes in real life, I can't =0A> think of a genera=
l behaviour that would make sense from the client side. In a =0A> shared re=
pository, you never know what revision number your next commit will =0A> ha=
ve.=0A> =0A> For a UI, I can envisage two useful ways to expose this functi=
onality: commit N =0A> empty revisions as a stand-alone operation, and comm=
it N empty revisions before =0A> the first revision loaded from a dump stre=
am. Obviously the former is =0A> sufficient; the latter is convenient but i=
s insufficient on its own unless we =0A> also give 'svnadmin load' a conven=
ient way to specify there is no dump =0A> stream to load.=0A> =0A> Stand-al=
one:=0A> =0A> =A0 svnadmin/svnlook commit-empty-revs N=0A> =A0=A0=A0 Commit=
 N empty revisions.=0A> =0A> As an option to 'svnadmin load':=0A> =0A> =A0 =
svnadmin load --prefix-empty-revs N=0A> =A0=A0=A0 First commit N empty revi=
sions.=0A> =0A> or, expressing a similar behaviour in a different way:=0A> =
=0A> =A0 svnadmin load --commit-first-loaded-rev-as X=0A> =A0=A0=A0 First c=
ommit enough empty revisions to make the first loaded revision=0A> =A0=A0=
=A0 be committed as revision number X.=0A> =0A> What should the author and =
log message be on the empty revs? I suppose these =0A> need to be optionall=
y specified, defaulting to blank?=0A> =0A> What should the date stamps be o=
n the empty revs? A thought: it seems cleaner to =0A> specify that they sho=
uld all have the same date stamp than that they =0A> do/don't/may all have =
different date stamps. (Imagine a future back-end in =0A> which we can crea=
te millions of 'virtual' empty revs in O(1) time and =0A> space as long as =
their rev-props are all identical.) The default for =0A> 'svnadmin load --p=
refix-empty-revs' without '--ignore-dates' =0A> ("ignore revision date stam=
ps found in the stream") should, I suppose, =0A> be that all the prefix emp=
ty revs have the same date stamp as the first revision =0A> loaded.=0A> =0A=
> - Julian=0A> =0A> [1] =0A> <http://www.timj.co.uk/2011/09/generating-empt=
ypadding-revisions-in-an-svn-dump/>=0A> [2] =0A> <http://stackoverflow.com/=
questions/7030041/can-i-create-a-subversion-repository-starting-at-another-=
number>=0A>=0A