summaryrefslogtreecommitdiff
path: root/docs
diff options
context:
space:
mode:
authornanbor <nanbor@ae88bc3d-4319-0410-8dbf-d08b4c9d3795>1998-05-11 19:09:03 +0000
committernanbor <nanbor@ae88bc3d-4319-0410-8dbf-d08b4c9d3795>1998-05-11 19:09:03 +0000
commit52888988f303ca13ed2ca36968c06ccce1987f75 (patch)
treebcb66aed95dc39bd54dfd10bb4d4e100f5f624bc /docs
parentfe7c39fc489fdc44611c370d19bae831d507110e (diff)
downloadATCD-52888988f303ca13ed2ca36968c06ccce1987f75.tar.gz
*** empty log message ***
Diffstat (limited to 'docs')
-rw-r--r--docs/CE-status.txt50
1 files changed, 26 insertions, 24 deletions
diff --git a/docs/CE-status.txt b/docs/CE-status.txt
index 48c351966a0..c7d9b9cb77a 100644
--- a/docs/CE-status.txt
+++ b/docs/CE-status.txt
@@ -2,30 +2,6 @@
[Current difficulties]
- [FILETIME]
- Win32's FILETIME format can be converted to struct timeval thru a
- simple conversion. However, the tricky part here is, FILETIME is
- the total 100ns elapse since _1601_ (why the heck did Microsoft
- choose this time?) whereas POSIX timeval gives you the
- accumulated time in us since 1970. This leads to data overflow
- in timeval struture which ACE uses internally for its
- ACE_Time_Value. This also confuses ACE_OS::ctime, which, under
- CE, is a home-brewed implementation, and corrupt the whole
- program image. I don't know of a way to deal with this yet
- except adjusting for the time difference between the two
- everytime a conversion is necessary. This, of course, doesn't
- seem to be a good choice.
-
- [ARGV]
- Is kind of falky. I use it to pass in command line arguments.
-
- [CE Toolkit]
- Using CE toolkit for C++ is simply a pain in the ass. It takes
- around 30 seconds to start up a debug session and 5+ seconds to
- single step thru a statement. It only allows remote debugging
- using the serial link but not the network. (BTW, currently, CE
- only support NE2000 compatible PCMCIA ethernet card.)
-
[Hints on writing portable code using ACE]
[Testing Non-Window based ACE programs]
@@ -88,6 +64,14 @@
[CE Toolkit issues]
+
+ [CE Toolkit]
+ Using CE toolkit for C++ is simply a pain in the ass. It takes
+ around 30 seconds to start up a debug session and 5+ seconds to
+ single step thru a statement. It only allows remote debugging
+ using the serial link but not the network. (BTW, currently, CE
+ only support NE2000 compatible PCMCIA ethernet card.)
+
[SH3 compiler]
* The SH3 compiler does not generate string literals when they
are used inside a function and when a precompiled header is
@@ -107,6 +91,24 @@
imported class to not use any virtual functions.
+[Tidbits]
+
+ [FILETIME]
+ Win32's FILETIME format can be converted to struct timeval thru a
+ simple conversion. However, the tricky part here is, FILETIME is
+ the total 100ns elapse since _1601_ (why the heck did Microsoft
+ choose this time?) whereas POSIX timeval gives you the
+ accumulated time in us since 1970. This leads to data overflow
+ in timeval struture which ACE uses internally for its
+ ACE_Time_Value. This also confuses ACE_OS::ctime, which, under
+ CE, is a home-brewed implementation, and corrupt the whole
+ program image. I don't know of a way to deal with this yet
+ except adjusting for the time difference between the two
+ everytime a conversion is necessary. This, of course, doesn't
+ seem to be a good choice.
+
+
+
[ACE File list]
Here is a trace of ACE files about their status on CE.
[Legend]