gmane.comp.sysutils.backup.fsvs.devel

[ANNOUNCE] FSVS 1.2.5 released

P.Marek — 2012-07-10T07:59:27

Hello everybody,

FSVS 1.2.5 is released.

The (only) change is a bug-fix for auto-props when explicitly specifying targets for
commit.


Please get the sources from

    https://freecode.com/projects/fsvs

(or from http://download.fsvs-software.org for the impatient) and/or hope for newer
packages.


Any feedback welcome!


Regards,

Phil

[fsvs-1.2.4] package build fails

Gunnar Thielebein — 2012-02-09T16:04:23

Hi Phil,

building the recent 1.2.4 version fails. I have seen there are changes in the
Makefile.in regarding apr libs.
After adding the libs the building still fails.

Snippet from error log is:


Think the dependencies are fullfilled e.g.:

I added the following to Makefile.in:

Do you have an idea what can cause this problem?

Regards,
Gunnar

------------------------------------------------------
http://fsvs.tigris.org/ds/viewMessage.do?dsForumId=3923&dsMessageId=2920145

To unsubscribe from this discussion, e-mail: [dev-unsubscribe< at >fsvs.tigris.org].

[ANNOUNCE] FSVS 1.2.4 released

P.Marek — 2012-01-17T17:25:32

Hello everybody,

I've released a new version of FSVS.


Changes since 1.2.3 are
 * Compilation fixes for MacOS 10.6; thanks, Thomas!
 * Added "password" option, as sent by Mark. Thank you!
 * Workarounds for gcc-4.5 and gcc-4.6 regressions. Thank you, Brian!
 * Compatibility with autoconf 2.68.


Please get the sources at

    http://freecode.com/projects/fsvs

and/or hope for new packages of your distribution (hint, hint ;)


As always: All ideas/bugs/questions are welcome, preferably to the mailing list(s).


Regards,

Phil

------------------------------------------------------
http://fsvs.tigris.org/ds/viewMessage.do?dsForumId=3923&dsMessageId=2909469

To unsubscribe from this discussion, e-mail: [dev-unsubscribe< at >fsvs.tigris.org].

Re: RHEL 5 / CentOS RPM Package for FSVS 1.2.2 (Re: [ANNOUNCE] FSVS 1.2.2 released)

David Fraser — 2011-07-26T06:05:25

To anyone concerned; as I said below I'm not maintaining the Fedora packages; the package has now been retired due to the lack of maintenance, so it's now up to anyone-who-wants-it to pick it up again...

----- Original Message -----
From: "David Fraser" <davidf< at >sjsoft.com>
To: "Farzad FARID" <ffarid< at >pragmatic-source.com>
Cc: "dev" <dev< at >fsvs.tigris.org>
Sent: Wednesday, February 16, 2011 8:26:43 AM
Subject: Re: RHEL 5 / CentOS RPM Package for FSVS 1.2.2 (Re: [ANNOUNCE] FSVS 1.2.2 released)

Hi Farzad

Sorry, I didn't ever reply to you on this. Are you still interested in maintaining FSVS on Fedora? I'd love to hand over to someone - you'd need to go through the standard procedures and I'm not sure what they are.
If not, is anyone else interested? I don't use Fedora actively any more and so it's not sensible for me to maintain it

Regards
David

----- Original Message -----
From: "Farzad FARID" <ffarid< at >pragmatic-source.com>
To: "dev" <dev< at >fsvs.tigris.org>, "David Fraser" <davidf< at >sjsoft.com>
Sent: Wednesday, A

Re: Compile errors on OpenBSD 4.8

Frode Haugsgjerd — 2011-06-26T11:05:19

Off cource I will!
Thank you for you patience.

Frode

------------------------------------------------------
http://fsvs.tigris.org/ds/viewMessage.do?dsForumId=3923&dsMessageId=2777294

To unsubscribe from this discussion, e-mail: [dev-unsubscribe< at >fsvs.tigris.org].

Re: Compile errors on OpenBSD 4.8

P.Marek — 2011-06-26T07:06:25

Ok, I'll add that.

 
Hmmm, so the HAVE_LOCALES detection doesn't work - because of the compiler 
error.
I'll add another test for that.


Thank you for your time and effort!

When I've fixed all that - may I ask you to run a test-checkout & compile, to 
see whether it works?


Regards,

Phil

Re: Compile errors on OpenBSD 4.8

Frode Haugsgjerd — 2011-06-25T20:41:34

Hi again Phil !

2011/6/25 Philipp Marek <philipp< at >marek.priv.at>

That worked, and adding this to configure also worked:
if [ `uname -s` = "OpenBSD" ]
then
        # OSX 10.6 - thanks, Florian.
        EXTRALIBS="-liconv"
        have_fmemopen=no
fi




Thanks, but I always do this when  recompiling.

I talked to a Mr. "NicM" on #openbsd< at >freenode, who identified a use of
iconv_open outside #ifdef HAVE_LOCALES in helper.c

Then i supected HAVE_LOCALES was not set, and commented out the ifdef
around  #include <iconv.h>
Now fsvs compiles.

Re: Compile errors on OpenBSD 4.8

P.Marek — 2011-06-25T18:23:39

Hello Frode!


...
Well, this should be seen during configure - HAVE_FMEMOPEN - and then just the
debug-buffer options should be disabled during compilation.

You could try --enable-release - this would remove all debug print calls (although the
function would have to be removed by the linker). But that shouldn't be necessary!


...
...
Well, this is a compile error; with FSVS it's a linker problem.

Perhaps this is an artifact of not running configure again after the update; please
re-run "make clean", "configure", and "make".


Regards,

Phil

Re: Compile errors on OpenBSD 4.8

P.Marek — 2011-06-25T09:13:09

Oh, sorry, that should be 2009, of course.

Re: Compile errors on OpenBSD 4.8

P.Marek — 2011-06-25T07:04:31

Hello Frode!

On Friday 24 June 2011 Frode Haugsgjerd wrote:
No problem, glad that this is solved.
 

And what is the real problem?
est_ops.c: In function 'ops__read_special_entry':
est_ops.c:1728: error: 'ENODATA' undeclared (first use in this function)
This one should be fixed since r2406, Dez 2003 ... now we're at r2455, so I 
wouldn't understand how this could affect you.

There's no ENODATA anymore in the FSVS sources ... Please double-check whether 
the checkout is current!

 
Yes, that would be one of the use cases ...


Regards,

Phil

RE: Re: Compile errors on OpenBSD 4.8

Frode Haugsgjerd — 2011-06-24T00:24:36

Hi, so far I've posted via "Discussions" on the web page, I can't find a subscribe button there. After some digging, I found that a email to dev-subscribe< at >fsvs.tigris.org does the trick. Sorry for my ignorance. It's been a while since I used tools like mutt and mailman :)

I found the page http://fsvs.tigris.org/source/browse/fsvs/ and assumed that trunk was the latest and greatest. But if you assume...

It seems like I distracted you with all my talk about trampolines. In svn trunk the -ftrampolines flag was missing. When I tried the 1.2 branch, the trampoline error was fixed.

My GCC version is as reported by "gcc -v" 4.2.1
This is the version that comes bundled with the OS.
I found an error in my first post, the latest system I've tested this on is OpenBSD 4.9

My real problems are with ops__read_special_entry 
 in est_ops.c 

Btw, I use fsvs on debian and really would like to be able to use it on our OBSD firewalls. I tailored the example scripts to my use, and exchanged the automatic nightly commit to a

RE: Re: Compile errors on OpenBSD 4.8

P.Marek — 2011-06-23T07:49:01

Hello Frode!


First of all - how about subscribing, so that I don't have to moderate your emails?
If you like I can do that for you, too.


Uh, the trunk is defunct.


I'm not sure what you mean - the trampoline error was gone on the 1.2 branch, or only
the trampoline error remained?


Are these when using the 1.2 branch?

What gcc version are you using? On the 48 page I see
   Gcc 2.95.3 (+ patches), 3.3.5 (+ patches) and 4.2.1 (+patches)

I cannot find anything about the "-ftrampoline" option ... Please add it in the
generated Makefile, and try compiling again.


Regards,

Phil

RE: Re: Compile errors on OpenBSD 4.8

Frode Haugsgjerd — 2011-06-22T19:58:29

Ok, forgot to mention that the version i compiled was trunk r. 2447

Tried the 1.2 branch, same result, but except the trampoline error was gone.

Attached ./configure --enable-dev-fake 2>&1 |tee configure.log

and make output

------------------------------------------------------
http://fsvs.tigris.org/ds/viewMessage.do?dsForumId=3923&dsMessageId=2772356

To unsubscribe from this discussion, e-mail: [dev-unsubscribe< at >fsvs.tigris.org].

Re: Compile errors on OpenBSD 4.8

P.Marek — 2011-06-22T19:27:37

Hello Frode!

...

Please try a svn checkout:

    http://fsvs.tigris.org/svn/fsvs/branches/fsvs-1.2.x/fsvs

because it has a few FreeBSD fixes that are not in a released tarball yet.

If we get OpenBSD 4.8 running, I'll cut a fresh release.


Regards,

Phil

Compile errors on OpenBSD 4.8

Frode Haugsgjerd — 2011-06-22T17:36:34

Hi, I'm trying to compile fsvs on OpenBSD 4.8, and I have some problems. http://fsvs.tigris.org claims that this is reported to work, but all i can find is a thread from 2006 that don't answer my questions.

First some info about my setup.

OpenBSD 4.8
From obsd packages:
autoconf-2.60
GNU Make 3.81
pcre-8.02p1
apr-1.2.11p4
subversion-1.6.15p0
gdbm-1.8.3p0
rsync-3.0.7
gcc 4.2.1

To satisfy gcc i had to write #include <stdlib.h>
in /usr/include/alloca.h , as obsd dont have this header.

Then gcc complained about trampolines, and i added 
CFLAGS  += -ftrampolines
to src/Makefile.in

Now comilation fails on:
     CC est_ops.c
est_ops.c: In function 'ops__read_special_entry':
est_ops.c:1728: error: 'ENODATA' undeclared (first use in this function)
est_ops.c:1728: error: (Each undeclared identifier is reported only once
est_ops.c:1728: error: for each function it appears in.)
gmake[1]: *** [est_ops.o] Error 1
gmake: *** [default-target] Error 2

Can anyone provide some pointers to what i should try next? I guess

Re: RHEL 5 / CentOS RPM Package for FSVS 1.2.2 (Re: [ANNOUNCE] FSVS 1.2.2 released)

David Fraser — 2011-02-16T06:26:43

Hi Farzad

Sorry, I didn't ever reply to you on this. Are you still interested in maintaining FSVS on Fedora? I'd love to hand over to someone - you'd need to go through the standard procedures and I'm not sure what they are.
If not, is anyone else interested? I don't use Fedora actively any more and so it's not sensible for me to maintain it

Regards
David

----- Original Message -----
From: "Farzad FARID" <ffarid< at >pragmatic-source.com>
To: "dev" <dev< at >fsvs.tigris.org>, "David Fraser" <davidf< at >sjsoft.com>
Sent: Wednesday, April 7, 2010 2:25:42 PM
Subject: Re: RHEL 5 / CentOS RPM Package for FSVS 1.2.2 (Re: [ANNOUNCE] FSVS 1.2.2 released)

Hi David,

Yes I'm interested in helping maintaining the package! Tell me what I 
should do in order to submit the package to Fedora.

  Regards,

Le 07/04/2010 08:50, David Fraser a écrit :

[ANNOUNCE] FSVS 1.2.3 released

P.Marek — 2011-01-17T10:35:14

Hello everybody,

FSVS 1.2.3 is released.

This is a really minor release; only one bug was fixed (property name
"svn:mtime" could get overwritten by using the longer FSVS_* environment
variables), but portability for OSX was improved.


Thank you Florian, Mike and Mark.


Get it while it's hot!
     http://freshmeat.net/projects/fsvs


Regards,

Phil

Re: discussion of fsvs setup in common binary packages

P.Marek — 2010-09-01T06:49:21

Hello Gunnar!

Well, this option should just control what changes are detected for files ...

So, if you look into ops__update_single_entry(), there the full path is already taken
via
    STOPIF( ops__build_path(&fullpath, sts), NULL);

If the option is set, you can just test the patterns and set the entry to FS_REMOVED, if
necessary ...



The problem is that it's not stored that an entry is removed - that is found (by getting
ENOENT on lstat()) on the next invocation of "status", "commit", or whatever.

So there isn't really a list to change ...
And BTW you'd have to send the "unlink" events to the repository, too. So just removing
from the list of known files isn't enough.


As long as the confusion is resolved ...


Regards,

Phil

Re: discussion of fsvs setup in common binary packages

Gunnar Thielebein — 2010-08-31T16:15:47

For files that are already in version control the "fsvs commit" is for sure a 
good place. But what about the files that weren't commited yet und you want to 
get rid from "fsvs st"?

My intention is to trigger the add/unversion operations somehow when the ignore 
list is changed via  "fsvs ignore load/prepend/append/at=". Where do you think 
the storm breaks loose in the teacup?


Sorry, to mix-up things with that question. I did a quick test for myself now 
and resulting did not care about ignore list changes.

Best,
Gunnar


------------------------------------------------------
http://fsvs.tigris.org/ds/viewMessage.do?dsForumId=3923&dsMessageId=2653740

To unsubscribe from this discussion, e-mail: [dev-unsubscribe< at >fsvs.tigris.org].

Re: discussion of fsvs setup in common binary packages

P.Marek — 2010-08-31T15:10:11

Hello Gunnar!


"fsvs ignore" is the wrong place - that's for editing the patterns.

As an option which is used on "commit" it's ok.

"sync-repos" is of course not needed - only the filenames have to be checked against the
filelist, and get added/unversioned as necessary.

Or it could be used with "add" or "unversion" ...


I'm sorry, I seem to have caused confusion with my remark about "sync-repos".
I meant that there's already a command starting with the letters "sync", and because
FSVS accepts abbreviations a new command like "sync-with-ignore" could wreck somebody
trying to use "sync-repos".
(Although, with one more thought, someone who does "sync-repos" must not care about his
WC list anyway.)


Regards,

Phil

Re: discussion of fsvs setup in common binary packages

Gunnar Thielebein — 2010-08-31T14:06:46

I fully agree. But the work begins if files needs to be sorted out because of 
privacy or bloating the repo etc. I think fsvs behavior can, because of its 
clueness and different use case, be different in that area. Comparing to svn the 
fsvs ignore list is also more centric then some sporadic propsets containing 
ignore rules.


I am also more for an option that changes the behaviour of "fsvs ignore".
What about something like "sync-with-ignore" defaulting to false? Would it be 
necessary or even suffice to run a "sync-repos"?


------------------------------------------------------
http://fsvs.tigris.org/ds/viewMessage.do?dsForumId=3923&dsMessageId=2653589

To unsubscribe from this discussion, e-mail: [dev-unsubscribe< at >fsvs.tigris.org].