Minutes, June 27, 2000


    I have spoken with Andrew Salem, President and CEO of the IEEE-ISTO
    (Industry Standards and Technology Organization) about bringing the
    Dwarf 2 specification under their umbrella. It appears to be designed
    for exactly what we are doing -- to provide services (publication,
    distribution, editing, etc.) for standards developed by an industry group.

    I believe that using the IEEE-ISTO would benefit the Dwarf 2 standardization
    effort significantly. This would provide a well known source for the
    documentation and establish its credibility as an industry standard.

    There is a cost for the IEEE-ISTO service -- probably on the order of
    $5000/year. At the meeting I asked that people discuss this with their
    corporate management and investigate obtaining commitments for financial

Issue discussion:

000519.2    Location list for data member   

    Proposal 000302.1 made changes to the definition of constant. This
    proposal permits both loc list references and offsets within an object
    to be described by constant values. A DW_FORM_data4 or _data8 constant
    is a location reference, while other constants are offsets into the object.


000531.2    Entry_PC attribute

    This proposal addresses the issue of finding the entry for a function
    which does not have a valid HIGH_PC/LOW_PC address, or where the entry is
    not at the LOW_PC address. A function which is not contiguous would not
    have high and low pc values defined.


000531.1    Add DW_AT_pc_ranges attribute

    This provides a mechanism for describing discontiguous scopes for blocks
    or functions. As described, several problems were identified. One is
    that the proposal would require assemblers or linkers to support LEB128
    data types, since generally compilers are unable to determine offsets
    within object files. In other places in Dwarf 2 where an assembler or
    linkers is expected to provide values, these values are the size of a
    machine address. It was felt that while the functionality proposed was
    desirable, requiring extensions to assemblers and linkers in order to
    obtain this functionality was undesirable.

    Withdrawn pending revision.

000405.1    Global types section

    This proposal provides for a section which would contain the names of
    global types which are defined in a program. This optional section
    is similar to the Pubnames section and contains the fully qualified
    unmangled names of classes, including namespace, defined in the program.
    Template type name issues, if such exist, are not addressed at this point.


000517.2    Line Number Table Is_Stmt

    Following the discussion at the last meeting, this proposal was withdrawn
    pending rewording. Since this was determined to be only editorial, and
    not to contain any substantive changes to the Dwarf 2 standard, further
    consideration by the committee is not required. Editor and proposer will
    work out satisfactory wording for the non-normative descriptive text.

000523.1    Reconsider 991118.1 re UTF-8

    Seven-bit ASCII is a subset of UTF-8, but common 8-bit ASCII representations
    for European characters, such as umlaut or cedilla, are not part of UTF-8.
    This proposal adds the DW_AT_use_utf8 flag to DW_TAG_compilation_unit which
    indicates that the name is in UTF-8, otherwise the name is in the system
    character set, which is unspecified.