[Bug] Export path is subject to peg revision parsing

4 views
Skip to first unread message

Osipov, Michael (SMD IT IN) via users

unread,
Jun 16, 2023, 3:46:02 AM6/16/23
to us...@subversion.apache.org
Folks,

I could have either found a bug in the handling of input or a document
insufficiency. We have received a downstream bug:
https://github.com/websvnphp/websvn/issues/189

It boils down to this command:
> # svn export -r 50 'file:///var/svn/Playground/me%40you.d/@' m...@you.d.r50
> svn: E200009: 'm...@you.d.r50': a peg revision is not allowed here

I fail to see here that PATH1 is subject to PEGREV parsing:
> # svn help export
> export: Create an unversioned copy of a tree.
> usage: 1. export [-r REV] URL[@PEGREV] [PATH]
> 2. export [-r REV] PATH1[@PEGREV] [PATH2]
>
> 1. Exports a clean directory tree from the repository specified by
> URL, at revision REV if it is given, otherwise at HEAD, into
> PATH. If PATH is omitted, the last component of the URL is used
> for the local directory name.

Verified with:
> # svn --version
> svn, version 1.10.8 (r1899510)
> compiled Jan 31 2023, 15:01:25 on amd64-portbld-freebsd12.4
and
> # svn --version
> svn, Version 1.14.2 (r1899510)
> übersetzt am Jun 16 2023, um 09:25:07 auf amd64-portbld-freebsd12.4

This works:
> # svn export -r 50 'file:///var/svn/Playground/me%40you.d/@' m...@you.d.r50@
> A m...@you.d.r50
> Exported revision 50.

Now the opposite:
> # svn export -r 50 'file:///var/svn/Playground/me%40you.d/@' m...@you.d.r50@99
> svn: E200009: 'm...@you.d.r50@99': a peg revision is not allowed here

this seems to be a safe bet to always append an @ to the PATH1:
> # svn export -r 50 'file:///var/svn/Playground/hallo.c@' hallo.c@
> A hallo.c
> Export complete.
> # ll hallo.c
> -rw-r--r-- 1 root wheel 104 2017-03-27 13:56 hallo.c

Any clarification is appreciated. JIRA doesn't contain anything useful.

Michael

Osipov, Michael (SMD IT IN) via users

unread,
Jun 16, 2023, 3:49:13 AM6/16/23
to us...@subversion.apache.org
Scratch that. My thread from five years ago is still valid:
https://lists.apache.org/thread/lonftwtj2kmnjf5mlp91jyxz9xlsgv3d

The issue sill persists. The doc improvement from Daniel Shahaf haven't
been implemented yet.

Michael
Reply all
Reply to author
Forward
0 new messages