Hi Scott,
I really appreciate your feedback, especially with respect to how
other contributors might react. It is important to stress to all
contributors and committers that I'm a deep believer in community work
and I look forward to learning from and working with all of you. I
warmly remember
Hi Taher,
I can understand your frustration, but this:
> This commit is wrong and bad on multiple levels. Please revert
is an unproductive and unreasonable request. Whether you want to or not,
you have to justify the request. Obviously the developer doesn't know what
is wrong or they
OK, I think nobody else cares because you indeed rarely run several OFBiz
instances on the same machine, apart in demos and maybe in development.
So I'll create a script specific to demos where I need to sometimes kill only
one of the 3 instances and not all.
Jacques
Le 27/06/2017 à 15:12,
I think terminateOfbiz should NOT be portoffset-aware nor should it
kill only one instance. terminateOfbiz is a convenience task to kill
all instances which is mostly used during development. That's the only
purpose it has. I believe it should NOT be used in any automation
suite or buildbot or
We still have a problem with terminateOfbiz: it's not portoffset aware.
So when you run it, it closes all OFBiz instances running on a machine (eg on
VM demo)
I guess we only need to improve terminateOfbiz by passing a portoffset parameter
I'll create a Jira and provide a patch for review
OK, the shutdown command is somehow our SIGTERM, makes sense
Jacques
Le 26/06/2017 à 08:45, Taher Alkhateeb a écrit :
I do not think we should send SIGTERM in the first place. The proper
termination of OFBiz is with the --shutdown command which cleanly stops
everything and calls the shutdown
I do not think we should send SIGTERM in the first place. The proper
termination of OFBiz is with the --shutdown command which cleanly stops
everything and calls the shutdown hook. The gradle task "terminateOfbiz"
should only be used if the shutdown command fails (last resort) as cleary
mentioned
Thanks Taher,
I indeed missed that kill is not an atomic operation. I reverted at r1799852.
Before reading your last message I wanted to propose to set a delay between the
2 operations.
So SIGTERM fails. I'd really like to send SIGTERM before killing the process and only kill it if it's
Oh and to prove the second scenario, just add a sleep for a few seconds
after the SIGTERM kill command and observe how the system will crash:
if (line ==~ /.*org\.apache\.ofbiz\.base\.start\.Start.*/) {
exec { commandLine 'kill', '-TERM', line.tokenize().first() }
Thread.sleep(5000)
}
Again,
I'm completing my answer for the record to show exactly where your code is
faulty and wrong just like in the other previous cases.
1- you have a copy-paste pattern in your code
2- you are calling the kill command twice immediately one after another
without checking whether SIGTERM killed the
But what's wrong? It's not clear to me, do you see it Michael?
Not clearly answering this question just adds more work on my side, I still
don't see why I should revert.
Jacques
Le 25/06/2017 à 13:11, Michael Brohl a écrit :
I's like to propose the following process to get this worked out
I's like to propose the following process to get this worked out without
endless ping-pong here on the mailing lists:
1. revert the change as requested
2. provide a Jira with patch and review/discuss it there (what's the
problem, how should it be solved etc.)
3. come to a conclusion and a
OK, I take this question as your refusal to revert and cooperate. I'm done
spending time here and I leave it up to the community to see if they want
incorrect code in the code base.
On Sun, Jun 25, 2017 at 1:34 PM, Jacques Le Roux <
jacques.le.r...@les7arts.com> wrote:
> Mmm but anyway, I
Good, now I kindly ask you to try not to take up _my_ time for _your_
mistakes by reverting instead of arguing. You refuse every time I ask for a
revert and end up doing it anyway but after taking up what little time we
have by arguing when there is so much work to do.
On Sun, Jun 25, 2017 at
Mmm but anyway, I remember now how I thought that.
In the 2nd "if", if the process was terminated, kill will just say that it
can't find the process and exec will return. Else the process will be killed.
What's wrong then?
Jacques
Le 25/06/2017 à 12:28, Jacques Le Roux a écrit :
Ah wait, I
Ah wait, I see line still contains "start" so will be executed twice anyway. OK
I'll improve that.
Jacques
Le 25/06/2017 à 12:18, Jacques Le Roux a écrit :
I don't revert without explanations on why I should revert. Sorry, but I don't
find your explanations clear.
My explanation, tell my
I don't revert without explanations on why I should revert. Sorry, but I don't
find your explanations clear.
My explanation, tell my what's wrong:
The 2 "if" blocks are executed sequentially for each line containing "a start"and ignore other lines. I did not change the previous logic, just
As usual, you refuse to revert because you don't understand the code and I
pay the price of spending my time explaining. I hope you will reconsider
this time consuming and non-cooperative behavior.
The quick version:
- copy and paste antipattern
- incorrect conditional checking leading to both
What makes you think it's wrong? I tested it locally using 2 background
instances and it cleaned worked.
I also tried with one standard instance (not in background). It works, and you
get this message
:ofbiz FAILED
FAILURE: Build failed with an exception.
* What went wrong:
Execution failed
This commit is wrong and bad on multiple levels. Please revert
On Sat, Jun 24, 2017 at 10:56 AM, wrote:
> Author: jleroux
> Date: Sat Jun 24 07:56:45 2017
> New Revision: 1799736
>
> URL: http://svn.apache.org/viewvc?rev=1799736=rev
> Log:
> No functional change
>
> Improves
20 matches
Mail list logo