You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Daniel Shahaf <d....@daniel.shahaf.name> on 2020/03/31 09:23:29 UTC

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

stsp@apache.org wrote on Tue, 31 Mar 2020 08:53 -0000:
> +++ subversion/trunk/subversion/libsvn_fs_fs/fs_fs.c Tue Mar 31 08:53:55 2020
> @@ -2343,3 +2344,168 @@ svn_fs_fs__info_config_files(apr_array_h
>  }  
> +
> +static svn_error_t *
> +ensure_representation_sha1(svn_fs_t *fs,
> +                           representation_t *rep,
> +                           apr_pool_t *pool)

Please add a docstring.

> +{
> +  if (!rep->has_sha1)
> +    {
> +      svn_stream_t *contents;
> +      svn_checksum_t *checksum;
> +
> +      SVN_ERR(svn_fs_fs__get_contents(&contents, fs, rep, FALSE, pool));
> +      SVN_ERR(svn_stream_contents_checksum(&checksum, contents,
> +                                           svn_checksum_sha1, pool, pool));
> +
> +      memcpy(rep->sha1_digest, checksum->digest, APR_SHA1_DIGESTSIZE);
> +      rep->has_sha1 = TRUE;
> +    }
> +
> +  return SVN_NO_ERROR;
> +}  
> +
> +static svn_error_t *
> +reindex_node(svn_fs_t *fs,
> +             const svn_fs_id_t *id,
> +             svn_revnum_t rev,
> +             svn_fs_fs__revision_file_t *rev_file,
> +             svn_cancel_func_t cancel_func,
> +             void *cancel_baton,
> +             apr_pool_t *pool)

Please add a docstring.

> +{

Cheers,

Daniel

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Nathan Hartman <ha...@gmail.com>.
On Wed, Apr 1, 2020 at 7:43 AM Daniel Shahaf <d....@daniel.shahaf.name> wrote:

> Thanks for addressing the commit review.
>
> However, I can't end this mail without saying that I feel your conduct
> in these threads has been wanting, even hostile.


This isn't a defense or attack on anyone, but we're in stressful times
right now and I can feel it everywhere I turn.

We're all volunteers here, doing what we can. Can we all please give each
other some slack?

Cheers,
Nathan

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Nathan Hartman <ha...@gmail.com>.
On Wed, Apr 1, 2020 at 9:08 PM Daniel Shahaf <d....@daniel.shahaf.name> wrote:

> Daniel Shahaf wrote on Thu, 02 Apr 2020 00:05 +00:00:
> > Stefan Sperling wrote on Wed, 01 Apr 2020 11:58 +00:00:
> > > If it makes you feel any better, I'd be happy to stop voluteering for
> the
> > > RM role, and let you or someone else worry about the release going
> forward.
> >
> > I won't dignify this petulant threat with a response.  Feel free to
> > rephrase yourself.  (And lest you mistake the previous sentence for
> [...])
>
> To be clear: I'm still interested in resolving this issue; however,
> I can't even begin to debug it unless you give me not only the error
> message but also the backtrace.
>
> I'm listening.
>
I think you're talking about the build repcache file system is corrupt
error with fsfs v6. If so, stsp fixed it a few hours ago but didn't
understand why the fix was necessary and afterwards Denis tracked that down
to an old bug/omission.

See:
https://lists.apache.org/x/thread.html/rba6b3c166a639f438778698a11e5491eeac7e0a849a378fc75fa7c60@%3Cdev.subversion.apache.org%3E

and the messages that follow.

If that's not the issue you're referring to, then apologies for the noise.
:-)

I hope you're well and keeping safe/healthy from this coronavirus. And that
applies to your family as well.

Cheers,
Nathan

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Stefan Sperling <st...@elego.de>.
On Thu, Apr 02, 2020 at 01:08:05AM +0000, Daniel Shahaf wrote:
> Daniel Shahaf wrote on Thu, 02 Apr 2020 00:05 +00:00:
> > Stefan Sperling wrote on Wed, 01 Apr 2020 11:58 +00:00:
> > > If it makes you feel any better, I'd be happy to stop voluteering for the
> > > RM role, and let you or someone else worry about the release going forward.
> > 
> > I won't dignify this petulant threat with a response.  Feel free to
> > rephrase yourself.  (And lest you mistake the previous sentence for [...])
> 
> To be clear: I'm still interested in resolving this issue; however,
> I can't even begin to debug it unless you give me not only the error
> message but also the backtrace.
> 
> I'm listening.

When we ask someone to fix something, that means we are always asking for
some amount of effort to be spent at the receiving end of our message.
And of course, there is also some effort spent in formulating the request
in the first place.
This in itself is not a problem. It is simply part of working in a group.

Problems happens when the perceived(!) effort at either side gets out of
balance. "Perceived effort" because I don't know much effort you really
spent reviewing my patch. All I see is the message you send me afterwards.

It is easy to scroll over a patch and find small inaccuracies and mistakes.
Especially when we have a project rule book which tells us what inaccuracies
to look uot for. When the book says things like "functions have docstrings"
that means any commit which doesn't conform to this rule can be criticized
by the book. But that doesn't always mean that we should. Context matters.

What our rule book cannot know is the how much effort we are willing to
put into conforming to it, and the trade-offs we make because of that.

So when I asked Denis to fix a buildbot issue, I didn't ask him to do
things which I would not be willing to do myself if necessary.

When you asked me to write and backport docstrings, I tried to communicate
to you that's a bit too much effort for me. And you told me to do it anyway.

When I point out that I don't believe docstrings are as important as other
things on the plate, that means I am close to my effort limit. If you then
push harder and harder to get the result you want, what natually happens
is that I get frustrated with the work we are trying to do together.

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Daniel Shahaf <d....@daniel.shahaf.name>.
Daniel Shahaf wrote on Thu, 02 Apr 2020 00:05 +00:00:
> Stefan Sperling wrote on Wed, 01 Apr 2020 11:58 +00:00:
> > If it makes you feel any better, I'd be happy to stop voluteering for the
> > RM role, and let you or someone else worry about the release going forward.
> 
> I won't dignify this petulant threat with a response.  Feel free to
> rephrase yourself.  (And lest you mistake the previous sentence for [...])

To be clear: I'm still interested in resolving this issue; however,
I can't even begin to debug it unless you give me not only the error
message but also the backtrace.

I'm listening.

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Stefan Sperling <st...@elego.de>.
On Thu, Apr 02, 2020 at 12:05:27AM +0000, Daniel Shahaf wrote:
> Stefan Sperling wrote on Wed, 01 Apr 2020 11:58 +00:00:
> > FWIW, "You are expected to fix it" could also be interpreted as hostile.
> > It's a direct order from you directed at me, telling me what I should do.
> 
> You committed a bug so I asked you to fix it, just like when Denis broke
> the build you asked him to fix it.
> 
> However, that was the ONLY thing I actually expected you to do.  The
> debugging suggestions were posted for whomever happened to have time to
> investigate the bug, not for you personally.  I apologize for my part in
> the Swiss cheese of that miscommunication (whatever it was; I don't see it).
> 
> > I would feel a lot better if you did work to improve upon existing work
> > and then showed me "this is how you could have done it, what do you think?"
> > instead of making suggestion after suggestion and expecting results from me.
> 
> I don't know whence you got the notion that you're entitled to commit
> bugs and expect me to fix them.  You might as well commit patches
> without log messages and then blame me for not writing them for you.
> 
> > If it makes you feel any better, I'd be happy to stop voluteering for the
> > RM role, and let you or someone else worry about the release going forward.
> 
> I won't dignify this petulant threat with a response.  Feel free to
> rephrase yourself.  (And lest you mistake the previous sentence for
> another "direct order", allow me to clarify that it isn't.  It's an
> invitation which you may accept or decline at your leisure and
> discretion.)
> 
> Daniel
> 
> P.S. I may not be able to respond in a timely manner for the next few days.

Enjoy your time off.

My offer still stands. When you come back you can pick RM for 1.14 if you
want to. And this isn't meant as a threat, it's an offer. I will step aside
if anyone in this community sees that I'm somehow doing it wrong and will
then contribute to the release by testing/signing etc. as usual.

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Johan Corveleyn <jc...@gmail.com>.
On Thu, Apr 2, 2020 at 2:06 AM Daniel Shahaf <d....@daniel.shahaf.name> wrote:
>
> Stefan Sperling wrote on Wed, 01 Apr 2020 11:58 +00:00:
> > FWIW, "You are expected to fix it" could also be interpreted as hostile.
> > It's a direct order from you directed at me, telling me what I should do.
>
> You committed a bug so I asked you to fix it, just like when Denis broke
> the build you asked him to fix it.
>
> However, that was the ONLY thing I actually expected you to do.  The
> debugging suggestions were posted for whomever happened to have time to
> investigate the bug, not for you personally.  I apologize for my part in
> the Swiss cheese of that miscommunication (whatever it was; I don't see it).

Daniel,

As I told you off-list (but perhaps that wasn't such a good idea after
all, to respond off-list): I don't sense any hostility from Stefan
here. Perhaps impatience and frustration that RM'ing 1.14 is taking
him more time and effort than expected, and that "a missing docstring
on a private static function" is holding it back (for a patch that he
tried to get in at the last minute (a noble effort BTW to give a new
contributor a timely result)), but hostility? Nah.

OTOH: "You are expected to ..." does not sound like a question to me,
it really sounds like an order. Also: it's not solely up to Stefan to
fix this (or any) bug, not even if he caused the bug / omission
himself (in this case, by shepherding another's patch). Anyone can
join the effort and fix others' bugs. Otherwise, we could hold any
contributor here forever responsible for any work they did (which
invariable contains bugs). I guess that's what Stefan meant with
"Wouldn't it be enough if you or someone else added docstrings on
trunk
at some convenient point in time in the future" and "I would feel a
lot better if you did work to improve upon existing work and then
showed me "this is how you could have done it, what do you think?"
instead of making suggestion after suggestion and expecting results
from me."

Putting myself in Stefan's shoes, I can certainly understand the
feeling of "a good deed never goes unpunished", which seems to go with
taking on the RM task (or for sticking out your neck for anything,
really).

I'm sorry I'm not doing anything significant these days myself. I'm
happy Stefan is doing the RM work, and trying to get another release
out the door. Just as I'm happy that you (Daniel) and others are
keeping an eye on things, reviewing and bringing up issues, and
helping where they can. Just ... try to be as supportive and
encouraging as you can for each other's efforts, thanks :-).

-- 
Johan

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Daniel Shahaf <d....@daniel.shahaf.name>.
Stefan Sperling wrote on Wed, 01 Apr 2020 11:58 +00:00:
> FWIW, "You are expected to fix it" could also be interpreted as hostile.
> It's a direct order from you directed at me, telling me what I should do.

You committed a bug so I asked you to fix it, just like when Denis broke
the build you asked him to fix it.

However, that was the ONLY thing I actually expected you to do.  The
debugging suggestions were posted for whomever happened to have time to
investigate the bug, not for you personally.  I apologize for my part in
the Swiss cheese of that miscommunication (whatever it was; I don't see it).

> I would feel a lot better if you did work to improve upon existing work
> and then showed me "this is how you could have done it, what do you think?"
> instead of making suggestion after suggestion and expecting results from me.

I don't know whence you got the notion that you're entitled to commit
bugs and expect me to fix them.  You might as well commit patches
without log messages and then blame me for not writing them for you.

> If it makes you feel any better, I'd be happy to stop voluteering for the
> RM role, and let you or someone else worry about the release going forward.

I won't dignify this petulant threat with a response.  Feel free to
rephrase yourself.  (And lest you mistake the previous sentence for
another "direct order", allow me to clarify that it isn't.  It's an
invitation which you may accept or decline at your leisure and
discretion.)

Daniel

P.S. I may not be able to respond in a timely manner for the next few days.

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Stefan Sperling <st...@elego.de>.
On Wed, Apr 01, 2020 at 11:43:05AM +0000, Daniel Shahaf wrote:
> Stefan Sperling wrote on Wed, 01 Apr 2020 11:51 +0200:
> > On Wed, Apr 01, 2020 at 09:34:39AM +0000, Daniel Shahaf wrote:
> > > Stefan Sperling wrote on Wed, 01 Apr 2020 10:31 +0200:  
> > > > On Wed, Apr 01, 2020 at 01:43:28AM +0000, Daniel Shahaf wrote:  
> > > > > Daniel Shahaf wrote on Tue, 31 Mar 2020 09:23 +00:00:    
> > > > > > Please add a docstring.    
> > > > > 
> > > > > I've gone ahead and added this to STATUS so it doesn't slip through any
> > > > > cracks.  Feel free to remove the -0 vote once it's been addressed (you
> > > > > needn't round-trip through me for this).    
> > > > 
> > > > Wouldn't it be enough if you or someone else added docstrings on trunk
> > > > at some convenient point in time in the future, and then nominated that
> > > > revision as a regular follow-up change if it's deemed important enough?
> > > >   
> > > 
> > > No, it wouldn't.  You committed two new functions without docstring.
> > > That's a bug in your commit.  You are expected to fix it.  
> > 
> > I am not questioning that functions need docstrings in principle.
> > I am just questioning the urgent need to backport such docstrings.
> > 
> > I would understand your objection if these were public API functions.
> > But these are local static functions, so the docstrings don't serve
> > the public. They will only be read by SVN developers.
> > 
> > Why is it not good enough if we put these docstrings on trunk without
> > also having to worry about them becoming part of the next release?
> > 
> > I am just suggesting that you already did enough by just asking for
> > the docstrings to be added on dev@ in response to my commit. That's
> > a perfectly reasonable request. I am only questioning the need to add
> > this request to STATUS also. From the RM perspective every little thing
> > in STATUS contributes to delaying the release.
> 
> Adding something to STATUS does not signify an "urgent need to backport"
> it and does not make it a release blocker.
> 
> Thanks for addressing the commit review.
> 
> However, I can't end this mail without saying that I feel your conduct
> in these threads has been wanting, even hostile.

Daniel, we are simply having some disagreements over priorities of things
affecting the release process. I don't intend to be hostile towards you.

FWIW, "You are expected to fix it" could also be interpreted as hostile.
It's a direct order from you directed at me, telling me what I should do.
I would feel a lot better if you did work to improve upon existing work
and then showed me "this is how you could have done it, what do you think?"
instead of making suggestion after suggestion and expecting results from me.

If it makes you feel any better, I'd be happy to stop voluteering for the
RM role, and let you or someone else worry about the release going forward.

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Daniel Shahaf <d....@daniel.shahaf.name>.
Stefan Sperling wrote on Wed, 01 Apr 2020 11:51 +0200:
> On Wed, Apr 01, 2020 at 09:34:39AM +0000, Daniel Shahaf wrote:
> > Stefan Sperling wrote on Wed, 01 Apr 2020 10:31 +0200:  
> > > On Wed, Apr 01, 2020 at 01:43:28AM +0000, Daniel Shahaf wrote:  
> > > > Daniel Shahaf wrote on Tue, 31 Mar 2020 09:23 +00:00:    
> > > > > Please add a docstring.    
> > > > 
> > > > I've gone ahead and added this to STATUS so it doesn't slip through any
> > > > cracks.  Feel free to remove the -0 vote once it's been addressed (you
> > > > needn't round-trip through me for this).    
> > > 
> > > Wouldn't it be enough if you or someone else added docstrings on trunk
> > > at some convenient point in time in the future, and then nominated that
> > > revision as a regular follow-up change if it's deemed important enough?
> > >   
> > 
> > No, it wouldn't.  You committed two new functions without docstring.
> > That's a bug in your commit.  You are expected to fix it.  
> 
> I am not questioning that functions need docstrings in principle.
> I am just questioning the urgent need to backport such docstrings.
> 
> I would understand your objection if these were public API functions.
> But these are local static functions, so the docstrings don't serve
> the public. They will only be read by SVN developers.
> 
> Why is it not good enough if we put these docstrings on trunk without
> also having to worry about them becoming part of the next release?
> 
> I am just suggesting that you already did enough by just asking for
> the docstrings to be added on dev@ in response to my commit. That's
> a perfectly reasonable request. I am only questioning the need to add
> this request to STATUS also. From the RM perspective every little thing
> in STATUS contributes to delaying the release.

Adding something to STATUS does not signify an "urgent need to backport"
it and does not make it a release blocker.

Thanks for addressing the commit review.

However, I can't end this mail without saying that I feel your conduct
in these threads has been wanting, even hostile.

Daniel

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Stefan Sperling <st...@elego.de>.
On Wed, Apr 01, 2020 at 09:34:39AM +0000, Daniel Shahaf wrote:
> Stefan Sperling wrote on Wed, 01 Apr 2020 10:31 +0200:
> > On Wed, Apr 01, 2020 at 01:43:28AM +0000, Daniel Shahaf wrote:
> > > Daniel Shahaf wrote on Tue, 31 Mar 2020 09:23 +00:00:  
> > > > Please add a docstring.  
> > > 
> > > I've gone ahead and added this to STATUS so it doesn't slip through any
> > > cracks.  Feel free to remove the -0 vote once it's been addressed (you
> > > needn't round-trip through me for this).  
> > 
> > Wouldn't it be enough if you or someone else added docstrings on trunk
> > at some convenient point in time in the future, and then nominated that
> > revision as a regular follow-up change if it's deemed important enough?
> > 
> 
> No, it wouldn't.  You committed two new functions without docstring.
> That's a bug in your commit.  You are expected to fix it.

I am not questioning that functions need docstrings in principle.
I am just questioning the urgent need to backport such docstrings.

I would understand your objection if these were public API functions.
But these are local static functions, so the docstrings don't serve
the public. They will only be read by SVN developers.

Why is it not good enough if we put these docstrings on trunk without
also having to worry about them becoming part of the next release?

I am just suggesting that you already did enough by just asking for
the docstrings to be added on dev@ in response to my commit. That's
a perfectly reasonable request. I am only questioning the need to add
this request to STATUS also. From the RM perspective every little thing
in STATUS contributes to delaying the release.

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Daniel Shahaf <d....@daniel.shahaf.name>.
Stefan Sperling wrote on Wed, 01 Apr 2020 10:31 +0200:
> On Wed, Apr 01, 2020 at 01:43:28AM +0000, Daniel Shahaf wrote:
> > Daniel Shahaf wrote on Tue, 31 Mar 2020 09:23 +00:00:  
> > > Please add a docstring.  
> > 
> > I've gone ahead and added this to STATUS so it doesn't slip through any
> > cracks.  Feel free to remove the -0 vote once it's been addressed (you
> > needn't round-trip through me for this).  
> 
> Wouldn't it be enough if you or someone else added docstrings on trunk
> at some convenient point in time in the future, and then nominated that
> revision as a regular follow-up change if it's deemed important enough?
> 

No, it wouldn't.  You committed two new functions without docstring.
That's a bug in your commit.  You are expected to fix it.

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Stefan Sperling <st...@elego.de>.
On Wed, Apr 01, 2020 at 01:43:28AM +0000, Daniel Shahaf wrote:
> Daniel Shahaf wrote on Tue, 31 Mar 2020 09:23 +00:00:
> > Please add a docstring.
> 
> I've gone ahead and added this to STATUS so it doesn't slip through any
> cracks.  Feel free to remove the -0 vote once it's been addressed (you
> needn't round-trip through me for this).

Isn't a discussion of docstrings for local functions in a backport
nomination too much distraction? Adding comments has no effect on code
correctness, and they would have to be added on trunk first anyway.

Higlighting bugs and missing things is of course important, but this is a
relatively small detail that could just be dealt with on the side.
From the RM point of view, every objection we add to STATUS takes up
reviewers' time, even if it's "just" a -0.

Wouldn't it be enough if you or someone else added docstrings on trunk
at some convenient point in time in the future, and then nominated that
revision as a regular follow-up change if it's deemed important enough?

Or do you actually need the docstrings to make sense of the code?
If that was the case, then why not vote -1?

Re: svn commit: r1875921 - in /subversion/trunk: subversion/include/ subversion/include/private/ subversion/libsvn_fs_fs/ subversion/svnadmin/ subversion/tests/cmdline/ subversion/tests/libsvn_fs_fs/ tools/client-side/

Posted by Daniel Shahaf <d....@daniel.shahaf.name>.
Daniel Shahaf wrote on Tue, 31 Mar 2020 09:23 +00:00:
> Please add a docstring.

I've gone ahead and added this to STATUS so it doesn't slip through any
cracks.  Feel free to remove the -0 vote once it's been addressed (you
needn't round-trip through me for this).