[Developers] Uppercase letters in model name

[Ian Taylor]

  I've had good experiences with google code 
pages, and my feeling is that we could use the one 
for ADMB much more.

If you want to edit or create a new wiki page, you 
have to be added as a "Committer" by one of the 
folks listed as "Owner" on the people page: 
http://code.google.com/p/admb-project/people/list

The existing wishlist wiki page is surely a small 
subset of the wishes that are out there. Whether 
this wiki page is the best forum to priorities and 
wishes for future development is a good question. 
It would certainly be good to have more people 
able to post their wishes somewhere. It's possible 
that if the wiki page were organized into 
categories it could handle being expanded without 
getting confusingly cluttered with ideas.

Yes, anyone can post issues and there are probably 
a number of issues that get worked out within 
these email lists that could be tracked better and 
recorded for posterity on the google code site.

It would also be great to see the source code 
tracking tools become used for the development 
version of the ADMB.
-Ian

On 11/10/2010 12:57 PM, Ben Bolker wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
>    I've posted at
>
>   http://www.math.mcmaster.ca/~bolker/misc/admb-notes.pdf
>
>    I said "a slightly nasty way", which isn't as bad as "the worst way"
> :-) ...
>
>    Can one/how does one add to/edit the wiki on googlecode
> <http://code.google.com/p/admb-project/w/list>  ?
> I see that there's a way to "enter a comment" on an existing page.  Can
> one edit pages, or add them? I guess I have to be a member?
> <http://code.google.com/p/admb-project/people/list>  says "Join this
> project by contacting the project owners, usually through participating
> in a mailing list." ... ?
>
>     I would be happy to put an edited version of my notes on the wiki, or
> to try to sort out which parts are still "issues" and post them on the
> "issues" page ...  (I see that I do have permission to do that -- but I
> will only do things that way if that is what the project owners recommend.)
>
>    cheers
>      Ben
>
>
> On 11/10/2010 11:35 AM, John Sibert wrote:
>> Path and case issues are longstanding. Although most of us have been
>> able to achieve livable work arounds, it might be worth the effort to
>> rationalize them a bit.  Arni is correct - putting spaces in paths is
>> risky behavior.
>>
>> It would be great to see Ben's issue list and compare it with what has
>> been going on over the last 2 years. I suspect we have been addressing
>> other issues (eg getin rid of "generic" code, porting to different
>> compilers, simplifying installation, ...). Perhaps Ben could resurrect
>> the document.
>>
>> John
>>
>> On 11/09/2010 12:17 PM, Ian Taylor wrote:
>>> I'll point out that Ben Bolker raised concerns related to this issue
>>> over a year ago, where my memory is that he described the behavior as
>>> something like "case sensitive in the worst way". Ben's document is no
>>> longer accessible at
>>> http://people.biology.ufl.edu/bolker/admb-notes.pdf, but I wonder how
>>> the past year's progress has stacked up against other issues that he
>>> identified back then.
>>> -Ian
>>>
>>> On 11/09/2010 01:57 PM, Arni Magnusson wrote:
>>>> Hi all,
>>>>
>>>> Weihai Liu just sent me an email, pointing out an ADMB issue when a
>>>> model name contains uppercase letters:
>>>>
>>>> * Create a directory containing two files, simple.tpl and simple.dat.
>>>>
>>>> * Rename the files to siMPle.tpl and siMPle.dat.
>>>>
>>>> * Build the model, this will result in a siMPle executable.
>>>>
>>>> * Run the model.
>>>>
>>>> ---
>>>>
>>>> This example highlights two issues:
>>>>
>>>> 1. Just typing 'siMPle' will not run the model in Linux, because the
>>>> model insists on looking for 'simple.dat' in lowercase. The user will
>>>> either have to rename the executable, or specify -ind siMPle.dat.
>>>>
>>>> 2. All the output files (simple.cor, simple.par, simple.std, ...) are
>>>> in lowercase, while the source code, executable, and input file
>>>> contain uppercase letters. This can cause problems when using
>>>> automated tools to archive and/or analyze pairs of input and output
>>>> files.
>>>>
>>>> These issues are minor nuisance, but it's probably easy to change the
>>>> behavior of ADMB so that the siMPle executable looks for siMPle.dat
>>>> and writes siMPle.* output files. This line from model7.cpp looks
>>>> relevant:
>>>>
>>>>    for (int i=1;i<=len;i++) adprogram_name[i]=tolower(adprogram_name[i]);
>>>>
>>>> Spaces in path is asking for trouble, but uppercase letters are
>>>> usually supported in statistical computing.
>>>>
>>>> Arni
>>>>
>>>>
>>>>
>>>> On Tue, 9 Nov 2010, Weihai Liu wrote:
>>>>
>>>>> Hi Arni,
>>>>>
>>>>> When using your admb.el file on ubuntu, the view report file or view
>>>>> estimate file or view point estimate menu will report error, "can
>>>>> not find the file", IF my tpl file had mixed Capital letter in its
>>>>> filename, see if my tpl is mixBayes.tpl, then the report file will
>>>>> be all small case mixbayes.rep, the same for .cor, .par etc, so they
>>>>> can not find it, because the filename had been changed by admb, but
>>>>> actually the files are there, it is the admb not consistent, the
>>>>> .htp, .o or .cpp are correct, but all others being forced to use all
>>>>> small case letter as filename. I only test it for non re.
>>>>>
>>>>> That will not be a problem for windows, and will not be a problem if
>>>>> the tpl filename is all small case letters, I just let you know to
>>>>> see if you can fix this either through change your admb.el file or
>>>>> let the admb fix it. The admb I am using is 64bit for linux.
>>>>>
>>>>> Thank you.
>>>>>
>>>>> Weihai Liu
>>>>>
>>>> _______________________________________________
>>>> Developers mailing list
>>>> Developers at admb-project.org
>>>> http://lists.admb-project.org/mailman/listinfo/developers
>>> _______________________________________________
>>> Developers mailing list
>>> Developers at admb-project.org
>>> http://lists.admb-project.org/mailman/listinfo/developers
>>>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.10 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAkzbBxwACgkQc5UpGjwzenOligCfcW7ii7IGKl3hTixfhFr0gYdz
> bPgAnRSBdycIohkbeqbGtsRocyRFlkve
> =1cQ5
> -----END PGP SIGNATURE-----