gnuplot development

Tracker: Patches

5 Refactor lf_push() - ID: 2889923

Last Update: Comment added ( sfeam )

Details:

This patch mangles several related changes. If only some of these changes
are accepted I will be happy to prepare a respective patch. They are
mangled because they are not quite commutative. :-(

The following things are changed:
* The if_depth is saved within lf_state_struct. This is actually a bug fix
for nested loads and evals.
* The call feature is largely factored out of load_file() into seperate
functions to clean the structure of load_file().
* The common code of eval_command() and load_command() is factored out into
lf_push() and lf_pop().
* The recursion detection for the eval command is reworked to check for the
same command string within the respective stack.
* The history_command() uses do_string() now that there is a recursion
detection.

Juergen

Submitted:

Juergen Wieferink ( wieferink ) - 2009-10-31 11:22

Priority:

Status:

Closed

Resolution:

Accepted

Assigned:

Nobody/Anonymous

Category:

None

Group:

None

Visibility:

Public

Comments ( 6 )

Date: 2009-12-20 21:35 Sender: sfeam Thanks. I've applied this to 4.5. If no problems crop up it can probably go into the 4.4 series also.
Date: 2009-12-20 10:26 Sender: wieferink PS: Sorry for the inconvenience. Juergen
Date: 2009-12-20 10:24 Sender: wieferink Yes it is. Honestly, I have updated the patches within my own repository long since (which is simple compared to publishing it...). The updated patch is attached. Juergen
Date: 2009-12-20 05:08 Sender: sfeam This patchset no longer applies cleanly, because in the interim there was a change in the token accounting, with num_tokens being replaced by (num_tokens+1) in several places. 2009-11-16 Juergen Wieferink <wieferink@freenet.de> * src/command.c (do_string_and_free) src/misc.c (load_file): Fix memory issues of patch 2009-10-30 (SF Patch #2898455 and Bug #2895620). I can apply the patch anyway after manual fix-up, but I am wondering if the same s/num_tokens/(num_tokens+1)/ change is needed by the code in the patch itself. Is it?
Date: 2009-11-04 13:22 Sender: wieferink The new patch is divided into several sections each of which copes with one of the above mentioned changes.
Date: 2009-10-31 14:57 Sender: wieferink Come to think of it, there could be done even more. lf_push() gathers a whole lot of scattered global variables. Some of the values (in current CVS only one: FILE *fp) belongs to the current recursion level, the others to the last. I think that these values could be better organized. Either keep all these values in a "current" lf_state_struct or at least organize them in substructs... Juergen

Log in to comment.

Attached Files ( 3 )

Filename	Description	Download
jw-refactor-2009-12-20.patch	Patch updated for 2009-11-16	Download
jw-refactor-2009-10-31.patch	Patch	Download
jw-refactor-2009-11-04.patch	New (demangled) patch	Download

Changes ( 6 )

Field	Old Value	Date	By
close_date	-	2009-12-20 21:35	sfeam
resolution_id	None	2009-12-20 21:35	sfeam
status_id	Open	2009-12-20 21:35	sfeam
File Added	355861: jw-refactor-2009-12-20.patch	2009-12-20 10:26	wieferink
File Added	349510: jw-refactor-2009-11-04.patch	2009-11-04 13:21	wieferink
File Added	348965: jw-refactor-2009-10-31.patch	2009-10-31 11:22	wieferink

gnuplot development

Tracker: Patches

Comments ( 6 )

Log in to comment.

Attached Files ( 3 )

Changes ( 6 )

About SourceForge

Find Software

Develop Software

Community

Help