| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 278.1 | You shouldn't create_repository more than once | TOOK::GUERTIN | Wherever you go, there you are. | Thu Aug 23 1990 20:30 | 18 |
| The mcc_mir_create_repository() routine is designed to be used once to
effectively "register" your repository with the MIR. From that point
on, the repository is known to MIR (saved in the mcc_mir_directory.dat
and mcc_mir_attribute.dat files). Presumably, you need to remove it
when you de-install your MM. In other words, why remove it? Just stop
using it.
The mcc_mir_get_repository() routine is used to return to you your
repository-ID which is used to access the repository you have
"registered" with the MIR.
I understand that from a development view-point, it is useful to have
a delete_repository routine, however from a product viewpoint, it
shouldn't be needed. In short, I believe we should have one (we do,
but it is not documented), for symetry as well as debugging. Maybe
for V1.1?
-Matt.
|
| 278.2 | I just call RMS | TOOK::GUERTIN | Wherever you go, there you are. | Mon Aug 27 1990 18:37 | 14 |
| Sorry. Reply .1 was for the wrong note (should have been for note 272).
I am just calling RMS. So I support what RMS supports. I pass to RMS
the "location" as the RMS file-name, and repository-name.DAT as the
RMS default-name. RMS should expand the file-name and fill in the
gaps with repository-name.DAT.
When you specify the location, do you specify it with the colon or
without? Is the logical name the same as the repository name?
I will also be trying some combinations to see if I can reproduce what
you saw.
-Matt.
|
| 278.3 | Is this a clue? | CLUSTA::YANG | | Tue Sep 11 1990 16:27 | 13 |
| Thanks for the answer of Note 272.
The location has colon in it, but not the same as the
instance-repository-name.
I believe that RMS takes the logical name and translates it because
if the logical is defined everything goes well. But a clue is that no
attribute-repository.dat is created if the logical is NOT defined. Is
that something causing you to bypass the creation of the
attribute-repository.DAT file?
Yah-Kong,
|
| 278.4 | Maybe, I will try it | TOOK::GUERTIN | Wherever you go, there you are. | Wed Sep 12 1990 17:33 | 18 |
|
Yes the colon may be causing the problem. I tried reproducing the
problem, but I could not. However, I tried it without a colon in the
location. I will try it again with the colon. Note, however that
what you pass in for the location is passed directly to RMS as the
filename (fab$l_fna field), that is to say, the string is not modified
in any way. What you pass in for the repository name gets .DAT appended
and is passed to RMS as the default file name (fab$l_dns field), so it
should still work, assuming RMS expands before filling the default
fields.
If the logical is not defined, you should get back either an RMS error
message, RMS$_DEV (error in device name or inappropriate device type
for operation), or an MCC error message MCC_S_LOGNOTDEF (logical name
not defined).
-Matt.
|
| 278.5 | no error if no logical | CLUSTA::YANG | | Thu Sep 13 1990 13:05 | 6 |
| The fact is that if no logical is defined, I do not get any error back.
And, the logical name (unexpanded name) was used to generate the
repository-name.DAT file under the default directory.
Yah-Kong,
|