I need to modify one procedure from within a package. I need to touch both the declaration and the implementation. As I am maintaining patch files for each modification, I would like the changes to be minimal.
Can I update the package with just the changed procedure ( if yes, how ? ) , or do I need to supply the complete package definition and implementation?
You need to replace the whole package specification and body - you cannot operate on just part of a package.
Just to contradict everyone else . . .
Technically you could do it - you could write something that would take in your patch file, retrieve the existing package source from the database (using USER_SOURCE), apply your patch, and then recompile the package using EXECUTE IMMEDIATE.
However, I don't think it would be a very good idea - patch based fixing becomes very difficult to keep track of, especially once multiple patches and multiple databases are involved. Putting the whole file into source control is a lot better - your patch should still be clearly visible.
If the patch is to a third-party package, consider wrapping it - so that everything is a straight call through except your patch. Or put your patch into a standalone package that calls the first one. There is still a danger that a change to the original package could be incompatible with your patch.
You cannot. As far as I remember, the only way to avoid referencing objects invalidation is not to touch your package declaration, and perform only CREATE OR REPLACE PACKAGE BODY.
Since the declaration changes, you might consider putting the new procedure into a new package, to avoid touching the existing one. Packages using the new version of the procedure must be adapted anyways, to reflect the change in the declaration (unless it's a new parameter with a default value).
If you're on Oracle 11g, and you want to minimise invalidations of other objects, make sure to place the new declarations at the end of the package spec.
Related
This is meant to be a FAQ question, so please be as complete as possible. The answer is a community answer, so feel free to edit if you think something is missing.
This question was discussed and approved on meta.
I am using R and tried some.function but I got following error message:
Error: could not find function "some.function"
This question comes up very regularly. When you get this type of error in R, how can you solve it?
There are a few things you should check :
Did you write the name of your function correctly? Names are case sensitive.
Did you install the package that contains the function? install.packages("thePackage") (this only needs to be done once)
Did you attach that package to the workspace ?
require(thePackage) (and check its return value) or library(thePackage) (this should be done every time you start a new R session)
Are you using an older R version where this function didn't exist yet?
Are you using a different version of the specific package? This could be in either direction: functions are added and removed over time, and it's possible the code you're referencing is expecting a newer or older version of the package than what you have installed.
If you're not sure in which package that function is situated, you can do a few things.
If you're sure you installed and attached/loaded the right package, type help.search("some.function") or ??some.function to get an information box that can tell you in which package it is contained.
find and getAnywhere can also be used to locate functions.
If you have no clue about the package, you can use findFn in the sos package as explained in this answer.
RSiteSearch("some.function") or searching with rdocumentation or rseek are alternative ways to find the function.
Sometimes you need to use an older version of R, but run code created for a newer version. Newly added functions (eg hasName in R 3.4.0) won't be found then. If you use an older R version and want to use a newer function, you can use the package backports to make such functions available. You also find a list of functions that need to be backported on the git repo of backports. Keep in mind that R versions older than R3.0.0 are incompatible with packages built for R3.0.0 and later versions.
Another problem, in the presence of a NAMESPACE, is that you are trying to run an unexported function from package foo.
For example (contrived, I know, but):
> mod <- prcomp(USArrests, scale = TRUE)
> plot.prcomp(mod)
Error: could not find function "plot.prcomp"
Firstly, you shouldn't be calling S3 methods directly, but lets assume plot.prcomp was actually some useful internal function in package foo. To call such function if you know what you are doing requires the use of :::. You also need to know the namespace in which the function is found. Using getAnywhere() we find that the function is in package stats:
> getAnywhere(plot.prcomp)
A single object matching ‘plot.prcomp’ was found
It was found in the following places
registered S3 method for plot from namespace stats
namespace:stats
with value
function (x, main = deparse(substitute(x)), ...)
screeplot.default(x, main = main, ...)
<environment: namespace:stats>
So we can now call it directly using:
> stats:::plot.prcomp(mod)
I've used plot.prcomp just as an example to illustrate the purpose. In normal use you shouldn't be calling S3 methods like this. But as I said, if the function you want to call exists (it might be a hidden utility function for example), but is in a namespace, R will report that it can't find the function unless you tell it which namespace to look in.
Compare this to the following:
stats::plot.prcomp
The above fails because while stats uses plot.prcomp, it is not exported from stats as the error rightly tells us:
Error: 'plot.prcomp' is not an exported object from 'namespace:stats'
This is documented as follows:
pkg::name returns the value of the exported variable name in namespace pkg, whereas pkg:::name returns the value of the internal variable name.
I can usually resolve this problem when a computer is under my control, but it's more of a nuisance when working with a grid. When a grid is not homogenous, not all libraries may be installed, and my experience has often been that a package wasn't installed because a dependency wasn't installed. To address this, I check the following:
Is Fortran installed? (Look for 'gfortran'.) This affects several major packages in R.
Is Java installed? Are the Java class paths correct?
Check that the package was installed by the admin and available for use by the appropriate user. Sometimes users will install packages in the wrong places or run without appropriate access to the right libraries. .libPaths() is a good check.
Check ldd results for R, to be sure about shared libraries
It's good to periodically run a script that just loads every package needed and does some little test. This catches the package issue as early as possible in the workflow. This is akin to build testing or unit testing, except it's more like a smoke test to make sure that the very basic stuff works.
If packages can be stored in a network-accessible location, are they? If they cannot, is there a way to ensure consistent versions across the machines? (This may seem OT, but correct package installation includes availability of the right version.)
Is the package available for the given OS? Unfortunately, not all packages are available across platforms. This goes back to step 5. If possible, try to find a way to handle a different OS by switching to an appropriate flavor of a package or switch off the dependency in certain cases.
Having encountered this quite a bit, some of these steps become fairly routine. Although #7 might seem like a good starting point, these are listed in approximate order of the frequency that I use them.
If this occurs while you check your package (R CMD check), take a look at your NAMESPACE.
You can solve this by adding the following statement to the NAMESPACE:
exportPattern("^[^\\\\.]")
This exports everything that doesn't start with a dot ("."). This allows you to have your hidden functions, starting with a dot:
.myHiddenFunction <- function(x) cat("my hidden function")
I had the error
Error: could not find function some.function
happen when doing R CMD check of a package I was making with RStudio. I found adding
exportPattern(".")
to the NAMESPACE file did the trick. As a sidenote, I had initially configured RStudio to use ROxygen to make the documentation -- and selected the configuration where ROxygen would write my NAMESPACE file for me, which kept erasing my edits. So, in my instance I unchecked NAMESPACE from the Roxygen configuration and added exportPattern(".") to NAMESPACE to solve this error.
This error can occur even if the name of the function is valid if some mandatory arguments are missing (i.e you did not provide enough arguments).
I got this in an Rcpp context, where I wrote a C++ function with optionnal arguments, and did not provided those arguments in R. It appeared that optionnal arguments from the C++ were seen as mandatory by R. As a result, R could not find a matching function for the correct name but an incorrect number of arguments.
Rcpp Function : SEXP RcppFunction(arg1, arg2=0) {}
R Calls :
RcppFunction(0) raises the error
RcppFunction(0, 0) does not
Rdocumentation.org has a very handy search function that - among other things - lets you find functions - from all the packages on CRAN, as well as from packages from Bioconductor and GitHub.
If you are using parallelMap you'll need to export custom functions to the slave jobs, otherwise you get an error "could not find function ".
If you set a non-missing level on parallelStart the same argument should be passed to parallelExport, else you get the same error. So this should be strictly followed:
parallelStart(mode = "<your mode here>", N, level = "<task.level>")
parallelExport("<myfun>", level = "<task.level>")
You may be able to fix this error by name spacing :: the function call
comparison.cloud(colors = c("red", "green"), max.words = 100)
to
wordcloud::comparison.cloud(colors = c("red", "green"), max.words = 100)
I got the same, error, I was running version .99xxx, I checked for updates from help menu and updated My RStudio to 1.0x, then the error did not come
So simple solution, just update your R Studio
I am developing a go package, which is a little bit complex and thus I want to organize the source code into multiple directories.
However, I don't want the users of the package to have to use too long imports. Anyways, the internal structure of the package isn't their concern.
Thus, my package structure looks so:
subDir1
subSubDir1
subSubDir2
subDir2
subSubDir3
...and so on. All of them have their exported calls.
I would like to avoid that my users have to import
import (
"mypackage/subDir1"
"mypackage/subDir1/subSubDir2"
)
...and so on.
I only want, if they want to use an exported function from my package, they should have access all of them by simply importing mypackage.
I tried that I declare package mypackage in all of the .go files. Thus, I had source files in different directories, but with the same package declaration.
In this case, the problem what I've confronted was that I simply couldn't import multiple directories from the same package. It said:
./src1.go:6:15: error: redefinition of ‘mypackage’
"mypackage/mysubdir1"
^
./src1.go:4:10: note: previous definition of ‘mypackage’ was here
"mypackage"
^
./src1.go:5:15: error: redefinition of ‘mypackage’
"mypackage/mysubdir2"
^
./src1.go:4:10: note: previous definition of ‘mypackage’ was here
"mypackage"
^
Is it somehow possible?
You should not do this in any case, as the language spec allows a compiler implementation to reject such constructs. Quoting from Spec: Package clause:
A set of files sharing the same PackageName form the implementation of a package. An implementation may require that all source files for a package inhabit the same directory.
Instead "structure" your file names to mimic the folder structure; e.g. instead of files of
foo/foo1.go
foo/bar/bar1.go
foo/bar/bar2.go
You could simply use:
foo/foo1.go
foo/bar-bar1.go
foo/bar-bar2.go
Also if your package is so big that you would need multiple folders to even "host" the files of the package implementation, you should really consider not implementing it as a single package, but break it into multiple packages.
Also note that Go 1.5 introduced internal packages. If you create a special internal subfolder inside your package folder, you may create any number of subpackages inside that (even using multiple levels). Your package will be able to import and use them (or to be more precise all packages rooted at your package folder), but no one else outside will be able to do so, it would be a compile time error.
E.g. you may create a foo package, have a foo/foo.go file, and foo/internal/bar package. foo will be able to import foo/internal/bar, but e.g. boo won't. Also foo/baz will also be able to import and use foo/internal/bar because it's rooted at foo/.
So you may use internal packages to break down your big package into smaller ones, effectively grouping your source files into multiple folders. Only thing you have to pay attention to is to put everything your package wants to export into the package and not into the internal packages (as those are not importable / visible from the "outside").
Inside your package source code, you have to differentiate your source directories by renamed imports. You can declare the same package mypackage in all of your source files (even if they are in different directories).
However, while you import them, you should give an induvidual names to the directories. In your source src1.go, import the other directories on this way:
import (
"mypackage"
submodule1 "mypackage/mySubDir"
)
And you will be able to reach the API defined in "mypackage" as mypackage.AnyThing(), and the API defined in mySubDir as submodule1.AnyThing().
The external world (i.e. the users of your package) will see all exported entities in myPackage.AnyThing().
Avoid namespace collisions. And use better understable, intuitive naming as in the example.
Yes, this is doable without any problems, just invoke the Go compiler by hand, that is not via the go tool.
But the best advice is: Don't do that. It's ugly and unnecessarily complicated. Just design your package properly.
Addendum (because the real intention of this answer seems to get lost sometimes, maybe because irony is too subtle): Don't do that!! This is an incredible stupid idea! Stop fighting the tools! Everybody will rightfully hate you if you do that! Nobody will understand your code or be able to compile it! Just because something is doable in theory doesn't mean this is a sensible idea in any way. Not even for "learning purpose"! You probably even don't know how to invoke the Go compiler by hand and if you figure it out it will be a major pita.
I am currently working on a very complex program that processes rows from an input table and has a huge number of possible outcomes for each record. Because of this I have a very large number of constants defined for the outcome messages. There is one success message for the record, but a multitude of possible warnings and errors.
My first thought was to define all of my constants for these messages at the package body level, but then I decided to move each constant to the procedure where it is used. I'm now second guessing that decision and thinking of moving everything back to package body level. What is the best way to define this many constants? Ease of maintainability is my ultimate goal for this program since it is so complex.
I think this is a matter of taste. In my application I put all error codes into an Error-Package. All main and commonly used constants I put into a separate package (without a package body).
Again, a matter of taste, but I tend to put a list of named constants at the package spec level rather than the package body so that they can be referenced by any portion of the application. If I ever want to change the error code that c_err_for_specific_reason_x uses, it becomes a single place to do so.
If I wanted to hide the codes and put them within the body I would have a get_error_code(p_get_error_name varchar) function that did the translation based on you passing a valid constant name.
I've done both on different projects, but tend towards the list over the function most times. I tend to use the function if it a table-driven source of the data.
It ... wait for it ... depends!
Since you currently define your constants in the package body, you don't need them to be publicly accessible outside the package. So defining them in a spec really doesn't buy you anything.
Here's is the rule I follow: Define constants within the smallest scope needed. So if a constant is used only within one procedure, define it in that procedure. If it is used within more than one procedure, define it in the body. If it is used elsewhere by code in other packages (or non-packaged SPs) but only when using a particular package, define it in the spec of that package. If it is used by other code for general use, put it in a separate spec of such general constants.
TASK:
Move all the functions and procedures in packages to the current Oracle schema. (you can imagine a case when you could need that, if not - take it like a challenge!)
QUESTION:
How can I read the functions/procedure "body" while they are in the package? I know that I can use all_source, dba_source and others to get the package body lines, but this means that I have to parse all those rows/strings - it should be an easier way. Isn't it?
If you have access to Toad, it does this very well.
Also, look at DBMS_METADATA package, specifically, the GET_DDL procedure.
Hope that helps.
Why exactly do you need this?
Are you just trying to execute the functions and procedures as if they were defined in your schema? If so, then invoker's rights may help.
Are you doing this for testing? If so, take a look at this answer: Is there a way to access private plsql procedures for testing purposes? (summary: use conditional compilation to optionally make functions and procedures public)
If you really need to break the packages down to functions and procedures you'll need to do it manually if you want to be 100% accurate.
There are many potential problems with just reading the source and trying to do it automatically. What about package variables, types, initialization, security (can every function be public?), procedures within procedures, duplicate names, wrapped source, etc.
I'm trying to replicate the code found in:
http://asktom.oracle.com/pls/apex/f?p=100:11:0::::P11_QUESTION_ID:59412348055
I did a copy and paste job. The package audit_pkg and the body compiled fine. But when I added the triggers the debug says "too many declarations for check_val" ...
Everything I've found says 10g supports overloading (or at least doesn't say otherwise).
Thoughts?
The procedure declarations in the package specification must match EXACTLY the declarations in the package body. This is where I usually encounter this error.
Is the column you're trying to use this package with a varchar2, number or date? If it's not, Oracle has to implicitly convert it to one of those three, and it won't know which one to use (and, therefore, which procedure to use). You may need to expand the package to handle more data types.