| Commit message (Collapse) | Author | Age | Files | Lines |
|\
| |
| | |
Windows crash in case of incorrect end statement
|
| |
| |
| |
| | |
Some Fortran compilers accept the END statement instead of the mandatory END TYPE. The code crashes on Windows as no correct END statement is found.
|
|/ |
|
|\
| |
| | |
Continuation lines and labels
|
| |
| |
| |
| |
| | |
Better handling of continuation lines when converting from fixed form Fortran to free form Fortran, especially handling the information in columns 733 and further (which is in fixed formatted Fortran always comment)
An `end` statement can start not only on lines with spaces in front of it but also when a line has a label or the `end` can be after a `;`.
|
|/
|
|
| |
Create an easy possibility to show the Fortran source as converted from fixed form to free form.
|
| |
|
|
|
|
| |
The Fortran scanner can give a message when the scanner is aborted, in here there is also the state, but in the form of a number. This patch adds a little bit more descriptive text.
|
|
|
|
|
|
|
|
|
|
|
|
| |
In fixed formatted Fortran code the numbers in the first positions of the comment were filtered out.
This is a regression on pull request #655
example where the number '1' disappears:
!> \details
!> 1.) First point
subroutine tst
end subroutine
|
|
|
|
|
| |
The backslash (\) is in Fortran not an escape character but was handled as such, this has been corrected.
Handling of the different Fortran string types ('...' and "...") made more consistent.
|
|\
| |
| |
| | |
into albert-github-feature/bug_enum
|
| |
| |
| |
| | |
All ENUMs are anonymous as they don't have a name defined in the Fortran standard.
|
|\ \
| | |
| | | |
Fortran improvements
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| | |
- adding NON_RECURSIVE
- better handling of missing PROGRAM in case of following module / subroutine in same file
- ignore (numeric)-labels in fixed source form
- adding support for
- TYPE IS
- CLASS IS
- CLASS DEFAULT
|
|/
|
|
|
| |
- handling of type / class statement especially in declarations in source browser and scanner
- color code import statement in source browser
|
|
|
|
|
| |
Super trivial typos
Some are in qtools/ which I know is a 3rd party dependency but as we know is now obsolete upstream. I reckon it wouldn't be much of an issue to merge neverthless
Tacked on several more commits
|
|
|
|
| |
Corrected handling of (local) variables as functions as well as handling of non Fortran variables used in Fortran code.
|
| |
|
|
|
|
|
|
|
|
|
| |
- FORTRAN: comment in subroutine argument list
This patch fixes the problem that comment lines in continuation parts stopped the continuation part.
Patch consists of 2 parts:
- correct setting of continuation characters for fixed form source code
- ignoring lines with just comment (based on initial fix proposed by angus-g, but extended for empty lines and comment lines that contain a continuation character)
|
|
|
|
| |
improve performance
|
|\
| |
| | |
Bug 700381 - error state 21 with fortran code (fixed format)
|
| |
| |
| |
| | |
Problem is caused by similar quotes inside quotes, in Fortran it is possible to "escape" quotes by doubling them.
|
|/
|
|
|
| |
This patch fixes a problem on Linux where in a number of cases doxygen crashed when an unnamed interface existed and the routine itself was in the same source file.
Furthermore the end line of routines is set correctly so inline source code in Fortran will be possible as well.
|
|
|
|
| |
The "value" attribute was already supported in the FORTRAN code browser, but not yet in the scanner.
|
|
|
|
| |
Signed-off-by: Adrian Negreanu <adrian.m.negreanu@intel.com>
|
|
|
|
| |
In case of a non generic interface in Fortran it happened that doxygen crashed due to the fact that the pointer for setting the endBodyLine does not exist. The line number is only defined for GENERIC interfaces and therefore it should not be set anyhow when ending a non generic interface.
|
|
|
|
| |
Small regression, line with 'C' (Capital C) was edited away, re-instated.
|
|
|
|
| |
According to the Fortran standard position 73 and further on a line are comment. Until now this was not considered.
|
|\
| |
| | |
Remove unused local and static global variables
|
| |
| |
| |
| | |
Remove unused local and static global variables. Variables have been identified by Understand (version 758) from Scitools.
|
| |
| |
| |
| | |
In case of a Fortran FUNCTION there was a difference between the handling of the source code lines for functions with and without the RESULT clause. This has been made uniform.
|
| |
| |
| |
| | |
For consistency also implemented for arguments.
|
|/
|
|
| |
Fortran functions can specify the return value by means of a variable in the RESULT attribute. Analogous to argument variables (@param) it is now possible to document also the return / RESULT variable (@returns).
|
|\
| |
| | |
Convert FORTRAN modules to namespaces
|
| |
| |
| |
| | |
This helps fix linking issues and ambiguity related to the difference between classes and modules in FORTRAN 2003+.
|
|/
|
|
| |
Fortran does not yet support the INLINE_SOURCES, with this patch (part of) this omission is solved.
|
|
|
|
| |
Added keywords contiguous and volatile
|
|
|
|
|
| |
Besides as as attribute the external can also be used as "type"
Previous commit (pull request 121) was reverted (pull request 122) due to warning message.
|
|
|
|
| |
recognized"
|
|
|
|
| |
Besides as as attribute the external can also be used as "type"
|
|
|
|
|
| |
In case the comment for Fortran arguments is empty or just contains @param [direction] <varname> or parts of these. The comment is is not shown and (depending on the configuration settings, a warning is given).
Also the detection of inconsistencies between intent attribute and documentation has been improved.
|
|
|
|
| |
Corrected documenation -> documentation
|
|
|
|
|
| |
Problem with variables with the name type versus type definitions.
type followed by = is recognized as not being a type definition instead of the use of a variable.
|
|
|
|
| |
For the determination whether the file is free or fixed formatted code just see lines starting with # as comment lines
|
| |
|
|\
| |
| |
| | |
https://github.com/albert-github/doxygen into albert-github-feature/bug_625601
|
| |
| |
| |
| |
| | |
The recognition of the type (free or fixed) of Fortran code is not reliable possible. A well known possibility as used with compilers as well is to specify the type of code by means of the extension.
With EXTENSION_MAPPING it is possible to select the type of Fortran code, when not explicitly set doxygen tries to guess the type of Fortran code.
|
|/
|
|
| |
Most problems were solved in version 1.8.5 but the underscore character (_) had not been incorporated, this is done with this patch.
|
| |
|
|
|
|
|
|
|
|
|
| |
- operations on current index and node (next(), prev(), last(), first()) have been removed.
- access to internal nodes has been removed.
- old QList has been renamed to QInternalList for use inside qtools only.
- added type safe compare, new, and delete operations (compareValues(), newValue(), deleteValue()).
- add compareValues also to QDict for consistency.
- changed doxygen's implementation to comply with the new QList and QDict interface.
|