Hello !
I do not know how whether I have my place in this discussion...
Several words, though.
I know I did not take very well one of your first messages, and I
progressively got the impression you were showing contempt toward this
work. Well, perhaps this word is not the best one, and I totally agree
with Robert when he mentions native speakers. This may have mattered.
Your first message was actually a request to change LP to MIP or MILP.
I know I very often do this, which is what you may call a mistake,
though as the difference between LP and MILP can always be deduced
from the context (or from the complexity of the problem mentionned), I
often forget it.
Well, I do not mind changing it, though I take it more as a stylistic
custom than something really important.
Actually, this patch being an algorithm for the H-minor problem, whose
review I expected to be long and tedious, I did not expect to receive
as a first comment a request to fix a typo. You also mentionned the
slowness of the algorith, which I had mentionned previously :
I understand this implementation is in many cases impractical, but I
thought odd that you would mention yourself its slowness even before
having tested it. I would have taken your remark much more differently
if you had only said "I tried to find this minor on this graph, and it
took me 3 hours", or given examples of running times. You did not
mention this, and I got the impression you were criticizing the
slowness without having even tested it yourself, just to check. I
still do not know whether you did, only you know as you did not
explicitely mention it.
Saying, just after this, that depending on its slowness it may be
useless to work on including this in Sage was a bit to aggressive for
me, when I had no proof you had not just taken a quick look at it.
Then again, english is not my native language, nor do I know what you
really meant in the end. This is just how I understood your message.
Later, you mentionned it may not scale, and I agreed with you, and you
sent another request :
"can you test it on 20-25 vertices ?"
I am sorry to have to say this, but I can obviously do it. As easily
as you would. It takes one line of Sage, namely ::
sage : graphs.RandomGraph(30, .3).minor(graphs.CompleteGraph(5))
Or something alike. As communications on the Trac server, because of
differences in our timezones, can sometimes take several hours, I
found a bit odd to be asked to type just one command, when you can
have done it yourself and obtained the answer immediately. It is a bit
like that when one is asked several times questions that are included
in the manual. I do not mind answering them, but at some point one is
expected to try by oneself. This also gave me the impression you still
had not applied and tested the patch, which confirmed what I already
thought of your first comments...
You also said "(by the way, "no K_4-minor" is equivalent to "treewidth
at most 2", so you can write another short function to test fro just
this...)". I think this is the place where, as Alex mentions it, a
reviewer is welcome to add a reviewer's patch. Very often, when
reviewing patches, I notice many small things that would take much
time to be explained over a Trac ticker, but which can be fixed in
several seconds by a short patch, none of which can really be
"refused" by the original author. It can come from fixing typoes to
adding an example in the docstrings, but in this situation it would
have required to add a very short function -- a one-liner. I would
have greatly appreciated to see you wrote a small patch based on mine
just to add a one-liner. It may not be long to do, but it is a perfect
proof to show in interest in the work being done. It is also a way to
quickly add a function to Sage when you find it useful and relevant.
Some time later, after these exchanges, you set the ticket to positive
review. I had, until then, absolutely no proof that you had read the
code. Actually, I hinted that you had not even applied the patch, and
none of your remarks had focused on the code.
I hope you will not think I did not invest time in the review process.
In this very ticket, I was asked to write a document explaining the LP
formulation better than the comments in the code would have, and I did
it. You can almost think I wrote a 8-pages long LaTeX document just
because of this very review. It took me -- a lot -- of time. Actually
longer than the time it took me to write the actual patch, when I
think about it.
Until now, I still do not know whether you read it, which was the
whole point of writing it.
I also read the following line from you :
////
Please fix this, otherwise I'll have to revert to "needs work" :) (I
wish I had such an efficient means to make my students work hard :))
////
Well, I now know you are not a PhD student. I am. I do not mind
working hard, I quite like both my job and the science I study, but I
take very badly the custom of some researchers to consider PhD student
as some unexpectedly useful form of animal life. One does not *make*
students work hard. One *asks* them. They can quit whenever they want,
this is the difference between them and slaves.
I said much more than "a few words", in the end. I will stop here as
it is the place where I have to apologize. I wrote all this to try to
make our misunderstanding clear, but I know that at some point I could
have avoided our conversation to lead a bit too far, which I failed to
do. So I apologize.
About the propositions you made, and I hope you will consider my
opinion as kindly as you would consider the opinion of any other, I do
not think anonymous reviewing would help. We're working together, and
there is nothing to earn by working on a free software. We are just
trying to build something together. Even if some, like me, sometimes
forget we have no reason at all to fight each other as we are all
working in the same direction, I believe as William said it that we
are the ones at fault.
With anonymous reviewers, the quality of review, and the whole
interest of working together would disappear.
Please receive my apologies, once more.
Nathann Cohen
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
