[PATCH 5/6] http: Avoid limit of retrying request only twice

Current code, after receiving HTTP_REAUTH, only retried
once, so couldn't do step 3 of the following sequence:

  1. We make a request; proxy returns 407, because we didn't give it a
     password. We ask for the password and return HTTP_REAUTH.

  2. We make another request; the proxy passes it to the actual server,
     who returns 401, because we didn't give an http password. We ask
     for the password and return HTTP_REAUTH.

  3. We make a third request, but this time everybody is happy.

Now we retry as long as we keep receiving HTTP_REAUTH, so the previous
sequence correctly completes.

Patch by Jeff King [off-list ref]

Signed-off-by: Nelson Benitez Leon <redacted>
Signed-off-by: Junio C Hamano <redacted>
---
 http.c |   11 +++++++----
 1 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/http.c b/http.c
index c87c66c..1468ec7 100644
--- a/http.c
+++ b/http.c

@@ -912,10 +912,13 @@ static int http_request(const char *url, void *result, int target, int options)
 static int http_request_reauth(const char *url, void *result, int target,
 			       int options)
 {
-	int ret = http_request(url, result, target, options);
-	if (ret != HTTP_REAUTH)
-		return ret;
-	return http_request(url, result, target, options);
+	int ret;
+
+	do {
+		ret = http_request(url, result, target, options);
+	} while (ret == HTTP_REAUTH);
+
+	return ret;
 }
 
 int http_get_strbuf(const char *url, struct strbuf *result, int options)

-- 
1.7.7.6

On Thu, May 03, 2012 at 06:40:28PM +0200, Nelson Benitez Leon wrote:

Current code, after receiving HTTP_REAUTH, only retried
once, so couldn't do step 3 of the following sequence:

  1. We make a request; proxy returns 407, because we didn't give it a
     password. We ask for the password and return HTTP_REAUTH.

  2. We make another request; the proxy passes it to the actual server,
     who returns 401, because we didn't give an http password. We ask
     for the password and return HTTP_REAUTH.

  3. We make a third request, but this time everybody is happy.

Now we retry as long as we keep receiving HTTP_REAUTH, so the previous
sequence correctly completes.

Patch by Jeff King [off-list ref]

We usually spell that as:

  From: Jeff King [off-list ref]

at the beginning of the email body (which lets am set the author
appropriately).

Other than that, the patch looks good to me (unsurprisingly :) ).

-Peff

On 05/04/2012 09:24 AM, Jeff King wrote:

On Thu, May 03, 2012 at 06:40:28PM +0200, Nelson Benitez Leon wrote:

quoted

[snip]

Now we retry as long as we keep receiving HTTP_REAUTH, so the previous
sequence correctly completes.

Patch by Jeff King [off-list ref]

We usually spell that as:

  From: Jeff King [off-list ref]

at the beginning of the email body (which lets am set the author
appropriately).

Are you saying the first line of the email body? isn't that for the
commit message first-line? I suppose you're not refering to the 'from
field' of the email as I would need alter identities in my email client
and found that cumbersome. Sorry for me not getting it and asking for
clarification.

On Fri, May 04, 2012 at 12:20:34PM +0200, Nelson Benitez Leon wrote:

On 05/04/2012 09:24 AM, Jeff King wrote:

quoted

On Thu, May 03, 2012 at 06:40:28PM +0200, Nelson Benitez Leon wrote:

quoted

[snip]

Now we retry as long as we keep receiving HTTP_REAUTH, so the previous
sequence correctly completes.

Patch by Jeff King [off-list ref]

We usually spell that as:

  From: Jeff King [off-list ref]

at the beginning of the email body (which lets am set the author
appropriately).

Are you saying the first line of the email body? isn't that for the
commit message first-line? I suppose you're not refering to the 'from
field' of the email as I would need alter identities in my email client
and found that cumbersome. Sorry for me not getting it and asking for
clarification.

Yes, that's what I am saying. When Junio applies the patch using "git
am", git will recognize these extra "pseudo-headers" at the top of the
body, remove them from the commit message, and override the email
headers with them.

It does this for exactly this case; you can leave your email headers
intact, but still provide attribution to someone else.

-Peff