#619 closed enhancement (fixed)
Use md5sum if ssl library is not found
Reported by: | Owned by: | ||
---|---|---|---|
Priority: | highest | Milestone: | |
Component: | nALFS - Back End (XML Handlers) | Version: | SVN |
Severity: | minor | Keywords: | |
Cc: |
Description
In case ssl library is not found, we could use md5sum executable.
Attachments (2)
Change History (12)
comment:1 by , 21 years ago
comment:2 by , 21 years ago
Owner: | changed from | to
---|
comment:3 by , 21 years ago
blocked: | → 613 |
---|
comment:4 by , 21 years ago
Could you move all non-handlers files to src/ (digest.c and md5.c). Although they are closely related to handlers, I think it's much cleaner this way, having only handlers themselves in src/handlers/. Should also make all that Makefiles changes unnecessary?
comment:5 by , 21 years ago
Status: | new → assigned |
---|
After thinking about it overnight I've decided that the best course of action is to create a lib/ subdirectory, that will contain the stuff used by other parts of nALFS. The Makefile in that subdirectory will build lib/lib.a.
This will also make the src/handlers/Makefile.in changes very minimal.
Additionally, I forgot to add some error messages to tell the user when the SSL library does not have the digest type they asked for, or if they asked for something other than MD5 without SSL linked in. Will post a new patch in a few minutes.
comment:6 by , 21 years ago
attachments.isobsolete: | 0 → 1 |
---|
comment:7 by , 21 years ago
qa_contact: | alfs-discuss@linuxfromscratch.org → alfs-log@linuxfromscratch.org |
---|
comment:9 by , 21 years ago
Priority: | lowest → highest |
---|---|
Resolution: | → fixed |
Status: | assigned → closed |
Commited to CVS.
comment:10 by , 20 years ago
Version: | CVS → SVN |
---|
I'd suggest a third alternative: just grab the md5 computation code from any of the dozens of GPL implementations and put it right into the download handler. I don't see any reason to be dependent on openssl (in fact, that's the only reason I even have openssl on my LFS system at this point), and using the md5sum binary is overkill. The MD5 algorithm is not going to change, and the existing code implementations are stable and bug free (and not very large).
And yes, I'm volunteering to do this if everyone agrees. If so, someone make me the owner of this bug and I'll get a patch made in the next few days (don't need CVS access, I'll just post a patch against CVS HEAD).