You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by "Jostein Chr. Andersen" <jo...@josander.net> on 2003/12/16 12:36:42 UTC
branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Hi,
We do now have the branch stabilizing week and it seems to work great.
I have a suggestion on how to use this week for making Subversion even
more stable (especially since we are reaching beta and soon 1.0):
-Checkout, compile and test the current release branch now!
It should not be any reason to see problems like "subversion don't
compile under myOS" and so on on this dev list _after_ a release.
Tests should be made on the diffrent OSs at once when the release branch
are made.
So, I think it would be nice if you can 'configure', 'make', 'make
check', 'make install', 'make davcheck' and 'make svncheck' on your OS/
distro/whatever with this packages installed:
-Berkeley DB 4.2.52
-A new SSH/SSL if possible (example: OpenSSH 3.7.1/OpenSSL 0.9.7c)
-Apache 2.0.48
-Neon 0.24.4
My system: Linux (LFS) with ReiserFS and all the latest packages above.
All my "make tests" went ok.
What do you think?
Jostein
--
http://www.josander.net/kontakt/ ||
http://www.josander.net/en/contact/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Jani Averbach <ja...@jaa.iki.fi>.
On 2003-12-16 13:36+0100, Jostein Chr. Andersen wrote:
>
> -Checkout, compile and test the current release branch now!
>
Everything went ok:
http://subversion.tigris.org/servlets/ReadMsg?list=svn-breakage&msgNo=6115
http://subversion.tigris.org/servlets/ReadMsg?list=svn-breakage&msgNo=6114
http://subversion.tigris.org/servlets/ReadMsg?list=svn-breakage&msgNo=6113
http://subversion.tigris.org/servlets/ReadMsg?list=svn-breakage&msgNo=6112
http://subversion.tigris.org/servlets/ReadMsg?list=svn-breakage&msgNo=6111
BR, Jani
--
Jani Averbach
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Ben Reser <be...@reser.org>.
On Tue, Dec 16, 2003 at 01:36:42PM +0100, Jostein Chr. Andersen wrote:
> It should not be any reason to see problems like "subversion don't
> compile under myOS" and so on on this dev list _after_ a release.
> Tests should be made on the diffrent OSs at once when the release branch
> are made.
Good idea. Did that on Mandrake 9.2 on amd64. Also built a pre-version
of my package and it looks good.
--
Ben Reser <be...@reser.org>
http://ben.reser.org
"Conscience is the inner voice which warns us somebody may be looking."
- H.L. Mencken
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Andy Whitcroft <ap...@shadowen.org>.
--On 16 December 2003 13:36 +0100 "Jostein Chr. Andersen"
<jo...@josander.net> wrote:
> We do now have the branch stabilizing week and it seems to work great.
> I have a suggestion on how to use this week for making Subversion even
> more stable (especially since we are reaching beta and soon 1.0):
>
> -Checkout, compile and test the current release branch now!
>
> It should not be any reason to see problems like "subversion don't
> compile under myOS" and so on on this dev list _after_ a release.
> Tests should be made on the different OSs at once when the release branch
> are made.
>
> So, I think it would be nice if you can 'configure', 'make', 'make
> check', 'make install', 'make davcheck' and 'make svncheck' on your OS/
> distro/whatever with this packages installed:
> -Berkeley DB 4.2.52
> -A new SSH/SSL if possible (example: OpenSSH 3.7.1/OpenSSL 0.9.7c)
> -Apache 2.0.48
> -Neon 0.24.4
>
> My system: Linux (LFS) with ReiserFS and all the latest packages above.
> All my "make tests" went ok.
>
> What do you think?
Great idea. Anyone up for making tarball with the right APR etc etc in it.
I am happy to compile and test on my platforms, but don't have the time to
go collect them all together. Think its quite simple a script or something.
Cheers.
-apw
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by kf...@collab.net.
Wonderful idea! This mail should be a regular part of the release
process, thanks for taking the initiative, Jostein.
-K
"Jostein Chr. Andersen" <jo...@josander.net> writes:
> We do now have the branch stabilizing week and it seems to work great.
> I have a suggestion on how to use this week for making Subversion even
> more stable (especially since we are reaching beta and soon 1.0):
>
> -Checkout, compile and test the current release branch now!
>
> It should not be any reason to see problems like "subversion don't
> compile under myOS" and so on on this dev list _after_ a release.
> Tests should be made on the diffrent OSs at once when the release branch
> are made.
>
> So, I think it would be nice if you can 'configure', 'make', 'make
> check', 'make install', 'make davcheck' and 'make svncheck' on your OS/
> distro/whatever with this packages installed:
> -Berkeley DB 4.2.52
> -A new SSH/SSL if possible (example: OpenSSH 3.7.1/OpenSSL 0.9.7c)
> -Apache 2.0.48
> -Neon 0.24.4
>
> My system: Linux (LFS) with ReiserFS and all the latest packages above.
> All my "make tests" went ok.
>
> What do you think?
>
> Jostein
>
> --
> http://www.josander.net/kontakt/ ||
> http://www.josander.net/en/contact/
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: dev-help@subversion.tigris.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Lübbe Onken <l....@rac.de>.
Branko Čibej wrote:
>>- Berkeley DB 4.0.2.50
>
> I don't believe that. :-)
You're right, I didn't even notice that. But do a right click on
libdb42.dll and check its properties. I copied the above string right
from there...
>>- Zlib 1.13
>
> Don't use this version of ZLib, even for private tests. It has a
> potential security hole.
Thanks for the info
Cheers
-Lübbe
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Branko Čibej <br...@xbc.nu>.
Lübbe Onken wrote:
> - Berkeley DB 4.0.2.50
I don't believe that. :-)
> - Zlib 1.13
Don't use this version of ZLib, even for private tests. It has a
potential security hole.
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Lübbe Onken <l....@rac.de>.
Jostein Chr. Andersen wrote:
> -Checkout, compile and test the current release branch now!
Checkout & compile are OK, but I get two test failures in the local tests.
System:
- Win2K SP4
- Berkeley DB 4.0.2.50
- Apache 2.0.48
- Neon 0.24.4
- OpenSSL 0.97b
- Zlib 1.13
Here are the failures of running "win-tests.py -r c:\tmp\junk"
FAIL: translate-test.exe 14: keep mixed line endings without repair flag
FAIL: commit_tests.py 22: commit from a dir with a longer name than the wc
If anyone is interested in more logs, please say so.
Cheers
-Lübbe
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Branko Čibej <br...@xbc.nu>.
Erik Huelsmann wrote:
>>Jostein Chr. Andersen wrote:
>>
>>
>>
>>>-Checkout, compile and test the current release branch now!
>>>
>>>
>>>
>>>
>>The Windows tests don't pass. translate-test 14 is failing regularly for
>>me. I've tracked it down to an open file handle that prevents us from
>>removing a file during the test. The handle leak is in
>>svn_subst_copy_and_translate.
>>
>>BTW, this is _not_ the same as the Windows access denied error.
>>
>>Just goes to show, we _must_ close files when we're done with them, like
>>I said.
>>
>>
>
>I disagree with that. We depend on APR to do that for us.
>
Well fine, but pray tell, how do you propose to delete a file on Windows
if its still open? The tests are failing on Windows because of this.
>We don't deallocate our memory when we are done with it either.
>
Files are not memory.
> If we were to free up all files
>and memory ourselves we would not need pool programming. There might be a
>problem with interaction between subversion's use of pools and Windows though.
>This could be either a problem in APR or in Subversion and we need to deal
>with that instead of working around it.
>
>
In this case, it's clearly a bug in the function which promises to close
the files it opened, but does a half-assed job of it.
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Erik Huelsmann <e....@gmx.net>.
> Jostein Chr. Andersen wrote:
>
> >-Checkout, compile and test the current release branch now!
> >
> >
> The Windows tests don't pass. translate-test 14 is failing regularly for
> me. I've tracked it down to an open file handle that prevents us from
> removing a file during the test. The handle leak is in
> svn_subst_copy_and_translate.
>
> BTW, this is _not_ the same as the Windows access denied error.
>
> Just goes to show, we _must_ close files when we're done with them, like
> I said.
I disagree with that. We depend on APR to do that for us. We don't
deallocate our memory when we are done with it either. If we were to free up all files
and memory ourselves we would not need pool programming. There might be a
problem with interaction between subversion's use of pools and Windows though.
This could be either a problem in APR or in Subversion and we need to deal
with that instead of working around it.
bye,
Erik.
--
+++ GMX - die erste Adresse für Mail, Message, More +++
Neu: Preissenkung für MMS und FreeMMS! http://www.gmx.net
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Erik Huelsmann <e....@gmx.net>.
[ snip ]
> You don't need an API change for that, let alone a change in the
> implementatino of the APR_DELONCLOSE flag. What you propose is in the
> same category as trying to change the open mode from readonly to
> readwrite without closing the file first.
>
> Instead of using APR_DELONCLOSE, you can register a pool cleanup that
> does the same thing, then remove it later if you've changed your mind.
> APR gives you all the mechanisms you need.
Ok, it seems a bit overkill to do that in this situation, but I'll have an
even closer look at pool programming to see about the pool cleanup.
Thanks for pointing it out.
bye,
Erik.
--
+++ GMX - die erste Adresse für Mail, Message, More +++
Neu: Preissenkung für MMS und FreeMMS! http://www.gmx.net
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Branko Čibej <br...@xbc.nu>.
Erik Huelsmann wrote:
>>Just goes to show, we _must_ close files when we're done with them, like
>>I said.
>>
>>
>
>BTW: you're looking at the code I am going to ask a function change for in
>APR. I thought to do that after 1.0; since there is only one instance we do
>this within Subversion I didn't think we have a very hard case.
>
>The thing is: we want a temporary file to be upgraded to a permanent one,
>but only after we reach a certain state. Therefore I was going to create the
>file with the APR_DELONCLOSE flag. After we reach the state we want to be in,
>we clear the APR_DELONCLOSE flag and close the file. It has been upgraded to a
>permanent one. If we don't reach the state we want to be in, we don't clear
>the flag and APR closes and deletes the file for us.
>
>This is all possible with the current APR implementation except for the
>Win32 one: it uses the (unclearable) system flag to create this behaviour,
>whereas all other implementations use one built into APR.
>
>
You don't need an API change for that, let alone a change in the
implementatino of the APR_DELONCLOSE flag. What you propose is in the
same category as trying to change the open mode from readonly to
readwrite without closing the file first.
Instead of using APR_DELONCLOSE, you can register a pool cleanup that
does the same thing, then remove it later if you've changed your mind.
APR gives you all the mechanisms you need.
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Erik Huelsmann <e....@gmx.net>.
> Just goes to show, we _must_ close files when we're done with them, like
> I said.
BTW: you're looking at the code I am going to ask a function change for in
APR. I thought to do that after 1.0; since there is only one instance we do
this within Subversion I didn't think we have a very hard case.
The thing is: we want a temporary file to be upgraded to a permanent one,
but only after we reach a certain state. Therefore I was going to create the
file with the APR_DELONCLOSE flag. After we reach the state we want to be in,
we clear the APR_DELONCLOSE flag and close the file. It has been upgraded to a
permanent one. If we don't reach the state we want to be in, we don't clear
the flag and APR closes and deletes the file for us.
This is all possible with the current APR implementation except for the
Win32 one: it uses the (unclearable) system flag to create this behaviour,
whereas all other implementations use one built into APR.
bye,
Erik.
--
+++ GMX - die erste Adresse für Mail, Message, More +++
Neu: Preissenkung für MMS und FreeMMS! http://www.gmx.net
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: branch 0.35.0: All my "make tests" are ok, how about you? ;-)
Posted by Branko Čibej <br...@xbc.nu>.
Jostein Chr. Andersen wrote:
>-Checkout, compile and test the current release branch now!
>
>
The Windows tests don't pass. translate-test 14 is failing regularly for
me. I've tracked it down to an open file handle that prevents us from
removing a file during the test. The handle leak is in
svn_subst_copy_and_translate.
BTW, this is _not_ the same as the Windows access denied error.
Just goes to show, we _must_ close files when we're done with them, like
I said.
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org