October 2007 - saga-rg - lists.ogf.org

saga core spec - final call
by Andre Merzky 11 Oct '07

11 Oct '07

Hi group(s), the deed is done, finally, and we incorporated all public comments back into the SAGA spec, and then some. And then some more. So, we would like to have a one week final call within the SAGA groups before we resubmit to the OGF editor. So, please, have a look at the document(s), and raise your voice if there is something you don't agree with. There are no open known issues at the moment, no TODO's and no FIXME's. But the text needs reviewing from a native (or good) English speaker, and also needs some formatting fixes. Shantenu and Thilo agreed to help with that, but more volonteers are VERY welcome! I attach two versions of the document: the clean version which is to be resubmitted, and a version which includes markup of the changes. The markup semantics is: red : removed green : fixed blue : added For large parts of the text (the verbatim sections), you'll find diff style markups + : added ! : fixed - : removed I'll try to come up with a detailed change log tomorrow or Friday, which should help you to understand the evolution of the document. BTW, the spec grew bigger again, but mostly because of added details, and not of added semantics. Cheers, Andre PS.: we know of course that a one week final call is short, but we would _really_ like to get the spec out of the door by OGF21. If you have concerns about the short call, please let us know! -- No trees were destroyed in the sending of this message, however, a significant number of electrons were terribly inconvenienced.

5 9

saga::file::read and saga::file::write etc.
by Hartmut Kaiser 09 Oct '07

09 Oct '07

Resending because of mailer problems... Hi all, While trying to implement the latest changes of the spec I have the following comments: 1. we added the option for a const correct buffer class to the spec, shouldn't the same be allowed for the iovec class (i.e. a const correct version of the iovec type) 2. I'ld suggest to change the saga::file::read and saga::file::write API functions as follows: saga::ssize_t read(saga::mutable_buffer, saga::ssize_t bytes = 0) saga::ssize_t write(saga::const_buffer, saga::ssize_t bytes = 0) The rationale behind this is that the buffer already contains a size, which by default could be used for the operation (if the 'bytes' parameter is zero). In C++ this would allow to simplify the code significantly. i.e. saga::file f(...); char cb[100]; f.read(buffer(cb)); // read 100 char's std::vector<double> v(20); f.read(buffer(v, 10)); // read 10 doubles Or std::vector<int> vi; // fill vi ... f.write(buffer(vi)); // write all elements of vi Etc. Regards Hartmut

2 2

Re: [SAGA-RG] saga::file::read and saga::file::write etc.
by Hartmut Kaiser 08 Oct '07

08 Oct '07

Hi all, > 2. I'ld suggest to change the saga::file::read and > saga::file::write API functions as follows: > > saga::ssize_t read(saga::mutable_buffer, saga::ssize_t bytes = 0) > saga::ssize_t write(saga::const_buffer, saga::ssize_t bytes = 0) > > The rationale behind this is that the buffer already contains > a size, which by default could be used for the operation (if > the 'bytes' parameter is zero). In C++ this would allow to > simplify the code significantly. i.e. > > saga::file f(...); > > char cb[100]; > f.read(buffer(cb)); // read 100 char's > > std::vector<double> v(20); > f.read(buffer(v, 10)); // read 10 doubles > > Or > > std::vector<int> vi; > // fill vi ... > f.write(buffer(vi)); // write all elements of vi The same is true for saga::stream::read and saga::stream::write, BTW. Regards Hartmut

1 0

Changelog
by Andre Merzky 08 Oct '07

08 Oct '07

Here is the promised Changelog. Cheers, Andre. ---------------------------------------------------------------- The SAGA Core API passed the public comment period. This changelog describes what changes have been applied to the document in reaction to the changes, and what additional editions took place. The purpose of the document is to help the reader to understand the evolution of the document, and to simplify the progress of the specification in the OGF document process. These changes are from version 1.0-RC.6 to 1.0-RC.7. - introduce a buffer class for uniform management of I/O buffers. - introduce a URL class for uniform parsing and handling of URLs and names. - radical overhaul of the permission model: ACLs are gone, and unix like permissions are introduced on all first class SAGA objects. - added PRECONDitions and PostConditions for all method calls. - added PERMissions for all method calls. - several SAGA interfaces now implement the task interface if they are used by classes which implement the task interface themself. - several SAGA interfaces now implement the permission interface if they are used by classes which implement permission task interface themself. - copy semantics for SAGA objects (shallow versus deep copy) has been clarified. - the lifetime of the default SAGA session has been clarified. - the SAGA class diagram has been updated. - the usage of URLs has been clarified. - byte ordering issues have been clarified. - relation of SAGA exceptions and middleware exceptions clarified. - fixed order/precedence of SAGA exceptions. - allow for nested exception in bindings (for implementations with late binding). - clarify usage of object types. - change saga::context to be more extensible and backend specific, while simplifying its attributes and initialization (set_defaults()). - clarify the formatting of user IDs for saga::context instances. - attributes cannot implicitely converted between vector and scalar attributes anymore - that was confusing. - simplified signature of find(). - changed the task interface: return values for async ops are essentially created at the point where they are used. - updated/fixed all state diagrams according to the new task model. - get_object() has been re-introduced on the task. - clarified that asynch object construction is a matter for the language bindings. - added verbose description for all enum types. - allow bindings to inherit the task container from native containers. - added support for the JSDL SPMD extension to the job description. - removed some unusable job description attributes. - clarified consistency semantrics of saga::logical_file instances. - clarify why no raw JSDL is supported. - clarify semantics of find() calls. - removed (unused) attribute interface from file. - unified semantics over the various packages: - open on object construction - exception semantics - close() semantics - state management - bitwise ORing of flags - several examples have been clarified. - countless spelling, grammar fixes and formatting fixes. ---------------------------------------------------------------- -- No trees were destroyed in the sending of this message, however, a significant number of electrons were terribly inconvenienced.

1 0

Re: [SAGA-RG] [saga-cvs] "SAGA-RG/Strawman TODO classes-alt.eps classes-alt.pdf classes.eps classes.fig classes.pdf classes.png job_states.eps job_states.pdf job_states_bes.eps job_states_bes.pdf saga_core.tex saga_core_attributes.tex saga_core_context.tex saga_core_design.tex saga_core_error.tex saga_core_examples.tex saga_core_files.tex saga_core_job.tex saga_core_logicalfiles.tex saga_core_monitoring.tex saga_core_namespaces.tex saga_core_permission.tex saga_core_rpc.tex saga_core_session.tex saga_core_s
by Ceriel Jacobs 01 Oct '07

01 Oct '07

Andre Merzky (merzky) wrote: > Update of /projects/SAGA-RG/Strawman > In directory asylum.cct.lsu.edu:/tmp/cvs-serv28572 > > Modified Files: > TODO classes-alt.eps classes-alt.pdf classes.eps classes.fig > classes.pdf classes.png job_states.eps job_states.pdf > job_states_bes.eps job_states_bes.pdf saga_core.tex > saga_core_attributes.tex saga_core_context.tex > saga_core_design.tex saga_core_error.tex > saga_core_examples.tex saga_core_files.tex saga_core_job.tex > saga_core_logicalfiles.tex saga_core_monitoring.tex > saga_core_namespaces.tex saga_core_permission.tex > saga_core_rpc.tex saga_core_session.tex saga_core_spec.tex > saga_core_stream.tex saga_core_tasks.tex states.eps states.pdf > stream_states.eps stream_states.pdf task_states.eps > task_states.pdf > Log Message: > - checked and fixed many examples, included Ceriels Java example > - added URL class (usage incomplete) > - removed flags from is_xxx calls in ns > - removed attrib interface from file > - fixed task model (added get_result and get_object) > - added implicit wait() on get_result > - fixed metric names according to the prescription of the monitorable interface > - checked metric types > - fixed class diagram > - added some default parameters > - clarified error reporting during object construction > - allow path normalization> > > > _______________________________________________ > saga-cvs mailing list > saga-cvs(a)cct.lsu.edu > https://mail.cct.lsu.edu/mailman/listinfo/saga-cvs > > _______________________________________________ > saga-cvs mailing list > saga-cvs(a)cct.lsu.edu > https://mail.cct.lsu.edu/mailman/listinfo/saga-cvs > - fix usage of IncorrectURL exception > - relaxed required lifetime of contexts (was impossible) > - removed seom redundant text in task model description > Andre. Hi Andre, I think saga_core_url.tex is missing from the CVS. "make" in Strawman fails. Ceriel

1 0