request for review and testing: default file name

Mikko Rasa tdb at tdb.fi
Mon Sep 10 08:42:26 PDT 2012


On 10.09.2012 17:30, Dirk Hohndel wrote:
>
> On Sep 9, 2012, at 11:05 PM, Henrik Brautaset Aronsen wrote:
>
>> Den 10.09.12 02:43, skrev Dirk Hohndel:
>>> This is reasonably well tested on Linux and somewhat tested on
>>> MacOS. Completely untested (beyond compliling) on Windows.
>>>
>>> I /think/ it implements a reasonable logic:
>>>
>>> if filenames are given, everything is like today.
>>> if no filename is given, it attempts to open the default file and sets
>>> it (even if the open failed) as the existing file name (so a first time
>>> user will get that as the default name when savin).
>>> The preferences dialog allows a change of the default file location.
>>>
>>> I'd especially appreciate feedback from Mac and Windows users - and of
>>> course general code review.
>>
>> Hi Dirk.
>>
>> I can't quite get it to work:
>>
>> 1) Open Subsurface
>> 2) In Preferences, verify that ~/Library/Application Support/Subsurface/henrik.xml is selected
>> 3) Log » Import » XML file » test1.xml
>> 4) Save
>> 5) Quit
>> 6) Re-open Subsurface
>> 7) No log is opened?!
>>
>> There's nothing written to ~/Library/Application Support/Subsurface/henrik.xml either (but the folder is created). And if I leave out 4), Subsurface doesn't ask me to save.
>
> I'll track this down. It's one of the thing that I tested in the process, but as usual some later "fix" may have broken it again.

An strace reveals that it's saving over the imported file.  For me, it's 
not even creating any directories in this case.

If I import dives from my dive computer instead, subsurface creates the 
directory ~/subsurface with 0600 permissions, and then fails to save the 
file because the directory lacks the execute permission bit.  The 
failure is not communicated to the user in any way.

After correcting the permissions, the file gets saved and then loaded on 
a subsequent startup.

I don't know if I like the correct behavior.  When I start an 
application in a blank state, do some changes, and tell it to save my 
work, I expect to be asked where to save it.  Perhaps just saving files 
somewhere is the way the industry is heading, with cloud computing and 
all, but until a more advanced FS comes along that lets me search for 
files based on the program that created them, I'd like to be in control 
of where the files are saved.

-- 
Mikko


More information about the subsurface mailing list